Latin faber 'smith', fabrica 'workshop'

Latin faber 'smith' < *dab-r-o- (traditional *dhabh-) derives from an onomatopoeic root *tap- ~ *dab- mimicking the hitting of metal¹. In Romance languages, the Latin word has descendants in Italian fabbro, Occitan faure, Old French fevre². Possibly also Basque harotz 'smith' reflects Italic *fabros.

The derivative fabrica 'workshop, forge' (but also 'art, trade') is the source of Spanish fragua 'forge': *frábica (metathesis) > frauga (syncope and lenition) > fragua. The Latin word is also the source of Catalan farga and French forge, the latter exported as a Wanderwort to several languages, including English.

______________________

¹ The IE etymology concoted by Pokorný, based on a resemblance to Armenian darbin, is a pile of rubbish.
² Hence orfèvre 'silversmith, goldsmith', a compound with or 'gold'.

Spanish urraca 'magpie'

The Catalan linguist Joan Coromines¹ derives Spanish urraca 'magpie (Pica pica)' from the female personal name Urraca², popular during the High Middle Ages in the Christian kingdoms of Northern Spain (Asturias/León, Castile, Navarre, Aragon). However, this hypothesis not only doesn't give us any insight about the etymology, but it also won't explain the initial consonant found in the dialectal variants burraca, furraca, hurraca /xuráka/, zurraca /θuráka/.

I'd link the Spanish word to Basque urra (B) 'expression to call hens and pidgeons; birds in general, especially hens and pidgeons' < *burra 'hen', with *bu > u³ like in Iberian. By contrast, Paleo-Basque, lacking of such a stop in its sound inventory, adopted it as /p/⁴, hence purra (G, HN, S, R, Z), furra (B, G, HN) 'expression to call hens', with the expressive variants tturra /cura/ (L, LN, Z), turra (L, LN). There's also Galician churra /tʃúra/ 'hen; expression to call hens'.

A diminutive form *burra-ka would be then the source of the forms burraca, urraca. In Spanish, the initial labial became a voiceless bilabial fricative *φ, variously reflected as /f/, /x/ or /θ/⁵. As in the case of most names of birds, the utlimate origin of this word would be onomatopoeic.

_____________________________

¹ J. Coromines (1973, 2008): Breve diccionario etimológico de la lengua castellana, p. 564.

² There's also the Galician variant Orraca.

³ Possibly through an intermediate stage *wu.
⁴ Cfr. Asturian utre 'vulture' ~ Gascon botre, butre > Basque putre (G), futre (HN).
⁵ Spanish *φ became aspirated as /h/ (except before /l, r, w/) at an early date, although the aspiration was later lost in most dialects. This is why the genuine form must have been hurraca.

Aragonese mardano, mardán 'ram' (updated)

Aragonese mardano, mardán 'ram'¹ is a pre-Latin word with cognates in Catalan mardà, marrà and Occitan marran, marre, marro, where it also means 'sturdy man'. The latter was borrowed into dialectal Basque marro (HN, L, LN) 'ram', barro (S, R) '1 year old ram'². Contrarily to some authors, in this word the original cluster -ɾδ- was assimilated to -ɾɾ- and not the other way around³. 

Basque has another word for 'ram' (also 'sheep' in northern dialects): ahari (L, LN), ãhãri (Z), aari (B, HN), adari (B), ari (B, G, S, R) < *anaɾi, from an earlier *a-maɾi with delabialization in Paleo-Basque. This protoform is cognate to Galician marón (diminutive marondo) 'male of animal (especially pig or bull)', regional Spanish marón⁴ and Gascon maro 'ram', a root found in Latin marītus 'male, husband' (likely an Etruscan loanword) and Hurrian mari(j)-annə 'chariot driver', which Starostin links to Caucasian *mo:rdɮV 'male'⁵.

In my opinion, this etymon could be also related to mardano and the like, although through a different substrate language.
_____________________________
¹ Also bardano (Echo), with denasalization of m-. See A. Kuhn (1935, 2008): El dialecto altoaragonés. p. 107.
² There's also regional Navarrese Spanish borro '1-2 years old lamb'.
³ A. Kuhn, op. cit., p. 116.
⁴ There're also the Spanish forms marote (diminutive), marueco, morueco, murueco.
⁵ I. Diakonoff & S. Starostin (1986): Hurro-Urartian as an Eastern Caucasian Language, §18.

Latin caesius 'light blue' (updated)

According to my amateur colleague Marco Moretti, Latin caesius 'light blue (also said of eye colour)' is a loanword from Etruscan caisie-, ceisi(e)-. These forms, together with caizna, ceizna, have exact counterparts in Baltic: Lithuanian gaĩsa-s 'glow, redness in the sky', Latvian gàiss 'air, wheather', gàišs 'bright, clear', gaisma 'light', etc.¹ In my opinion, these words are indicative of prehistoric contacts between Tyrrhenian (the language family to which Etruscan belongs) and a linguistic continuum from which emerged Baltic and Thracian². 

Also Lithuanian gaidrùs 'bright, clear (said of wheather)' and Greek phaidrós 'shining, bright, cheerful' belong here, and Duridanov reconstructs Thracian *gaidrus on the basis of the personal name Gaidre(a)s³. For these forms, an IE protoform *g^wai- 'light, bright'⁴ can be reconstructed, with regular delabialization of the labiovelar in "satem" languages. In my opinion, this is a loanword from Caucasian *ʔVqq^wo-ji- 'white, light, blueish', in turn a derivative of *=eqqwA 'yellow'.

From the same Caucasian source it also comes a different IE protoform *koi-t^ʔ-, found in Lithuanian skaidrùs 'clear, bright', skáistas, skaistùs 'bright', Germanic *xaid-u- > Gothic haidus 'way, manner', *xaid-a- > Old Norse heiδ 'clear (said of weather)', heiδr 'clear, cheerful', *xaid-ra- > Old High German heitar, German heiter 'cheerful', and to which some IE-ists link the Latin word⁵.

________________________________ 

¹ As Etruscan didn't possess voiced stops, c /k/ would be the rendering of a former /g/.

² This continuum was broken by the irruption of people from the Pontic Steppes, whose languages interacted with the autochthonous ones and gave raise to Dacian (a close relative of Indo-Iranian) as well as Greek and its nearest relatives (Macedonian, Phrygian, Armenian).

³ I. Duridanov (1969): Die Trakisch- und Dakisch-Baltischen Sprachbeziehungen, p. 75.
⁴ Ortodox reconstructions are *g^whai- and *g^wheH2i-. See R.S.P Beekes (2010): Etymological Dictionary of Greek, p. 1544.
⁵ M. de Vaan (2008): Etymological Dictionary of Latin and the Other Italic Languages, p. 82. 

Aragonese paniquesa 'weasel' (updated)

Aragonese paniquesa, Gascon panquèra, panquèsa 'weasel (Mustela nivalis)' is a diminutive form (with the Romance suffix -ella) which underwent a folk etymology from pan y queso 'bread and cheese'¹. This word shows the idiosyncratic treatment of -ll- in Pyrenaic², similar to the one of West Asturian, which has a voiceless retroflex africate [tʂ].
 

This is an Eurasiatic root found in Eskimo-Aleut *paniɣ 'daughter', Altaic *p^hjun[e] ‘a small wild animal’ (Tungusic *pün´- 'jerboa, flying squirrel, mole; weasel; hedgehog', Mongolian *hünegen 'fox', Turkic *enük (~ *ünek) 'young of a wild animal, puppy'), which I'd link (through labialization of the initial labiovelar) to Caucasian *ɦnǝ:qq’wǝ: (~ *ɦqq’wǝ:nǝ:) 'mouse, rat'³, Yeniseian *ku:n´ (~ g-) 'wolverine', Balto-Slavic *keun- 'marten'.

Possibly also related are Latin cunīculus (a Paleo-Hispanic loanword) and dialectal Basque untxi (HN, R), entxe (HN) 'rabbit', all them diminutive forms.

___________________________

¹ Hence the Basque calque ogigaztae (B) 'weasel', from ogi 'bread' and gaztae 'cheese'. For other names of 'weasel' in Basque, see here.

² An extinct Romance language spoken in the High Middle Ages and which gave loanwords to Aragonese and some Pyrenaic Gascon varieties, especially Bearnese. 

³ The metathesized variant is related to Uralic *n´ukk- 'fox', Dravidian *nakk- 'fox, jackal'.