On lenition, I have a problem with the GGBC rule that masculine nouns in the npl and gs lenite a following noun in the genitive (an indefinite genitival qualifier). Na sagairt pharóiste and an tsagairt pharóiste would be what the rule implies. This is from my file. Maybe An Lon Dubh knows more?

a) Lenition after a noun in the nominative plural ending in a slender consonant
It is difficult to find a large number of relevant phrases in traditional literature, and it may be that the rule in GGBC stating that all such indefinite qualifiers should be lenited may merely be intended to generate a simple rule for learners rather than to capture native usage. Fir chínn riain is a frequently encountered phrase that exhibits the rule appropriately. Yet Nua-Chorpas na hÉireann1 shows that Pádraig Ó Cíobháin had buidéil fíona in An Grá Faoi Cheilt and Seán Ó Lúing had buidéil branda in Gach Orlach de mo Chroí—Seoirse Mac Tomáis, both being native speakers of Kerry Irish. Another Kerry native, Maidhc Dainín Ó Sé, has na buidéil phórtair in Chicago Driver. It would be tempting to seek to impose a contrast between {buidéil fhíona}, “wine bottles”, and buidéil ǁ fíona, “bottles of wine”, but then Amhlaoibh Ó Loingsigh has leabhair-bhuidéil bhrannda ’na ndiaig mar dhig. Lothaill ǁ práis (where lothall is Ua Laoghaire’s form of logall, “socket”) is found repeatedly in Ua Laoghaire’s Bible manuscripts (e.g. Exodus 26:37). It may be that both lenition and non-lenition are acceptable with a minor distinction in nuance (adjectival vs. description of contents/material), with either acceptable in many contexts.
That the rule of lenition in the genitive after a plural noun ending in a slender consonant is not absolute was suggested by O’Nolan in his New Era Grammar (113), where he claimed that “dynamic aspiration” can create minor differences between phrases. He cited airm ǁ cogaidh, “war weapons”, and {airm chogaidh}, “weapons for a war”. This appears to set up a minimal pair where one phrase has a genericised noun that is not indefinite (cogadh<an cogadh) and the other has an indefinite noun. O’Nolan did not state whether he had heard such a distinction being made in native speech. In fact, airm ǁ cogaidh is frequently encountered in Muskerry literature, and must be accepted as the correct phrase in the dialect, whereas the related phrase {airm chosanta} has lenition. This is perplexing, but it could be argued that verbal nouns such as cosaint lend themselves to essentially adjectival usage. It remains unclear whether there were ever really minimal pairs of the sort claimed by O’Nolan.
Another strong possibility is that an original distinction between adjectival usage (with lenition) on the one hard, and generic usages and descriptions of contents or material on the other (without lenition) underlies the collocations attested, but that once a given collocation has become accepted as a well-known phrase, then it becomes fixed as such. From this point of view, buidéil ǁ fíona could be found whether “wine bottles” or “bottles of wine” were intended, a usage derived from the frequency with which “bottles of wine” are referred to. If “dynamic lenition” be accepted as a concept, it may be that the indefinite phrase lothaill ǁ práis would be found on first mention, but where the phrase is definite, na {lothaill phráis} might be found (as once the existence of brass sockets is already established, there is a tighter connection within the phrase when the brass sockets already known to exist and to be made of brass are subsequently referred to with the definite article). Other examples I have are osáin ǁ práis (“greaves of brass on his legs” in 1 Samuel 17:6); na corcáin práis (“the pots of brass” in 4 Kings 25:14); and scáláin ǁ práis (“a brazen scaffold” in 2 Paralipomenon 6:13). Na corcáin práis, if not a slip of the pen, might show that there is only a nugatory difference in nuance between “the brazen pots” and “the pots of brass”, and even definite phrases of this type could be phrased either way.
The plural noun na sagairt paróiste is not specifically attested in Muskerry literature (although a subsequent paragraph will show that the genitive singular an tsagairt paróiste without a lenited p is). Na sagairt pharóiste would be the form recommended by grammars such as GGBC. If the p were found unlenited, it might be held to show that the second noun was not an indefinite, but rather a genericised noun. Once again, a tiny potential difference in nuance may be admitted, with na sagairt pharóiste also correct if found.

I use ǁ to show that non-lenition ruptures the unity of phrases like lothaill ǁ práis. They become "sockets of brass", not "brass sockets", where brass is conceptually separate. And where there is lenition, I use the transcription {lothaill phráis} to show how lenition binds the phrase together.

Maidhc Dainín Ó Sé seems to stick to the rule always, as far as I have checked. Ignoring the stock/frozen mythological phrase "Na Fir Bolg".

Croí Cainte Ciarraí adheres to it pretty strictly as well, but not always.

For example:
Airm mhóna

but:
Airm feirmeora

Regarding another point you mentioned: "an tsagairt paróiste" without the lenition expected is found in two novels of Maidhc Dainín (Idir Dhá Lios and Lili Frainc). This can be a case of a "frozen nominative" for a common phrase. Essentially where parts of the phrase are treated like the nominative due to common use. Similar to "leis an tAthair X". However it could also be simply the fact that genitive mutations for adjectives are on the decline in native speech as mentioned in An Teanga Bheo: Corca Dhuibhne p.15. I'm not confident of either of these explanations though in the case of the older authors you mention.

I'll try to look for other examples where the adjective is attested in the both the lenited and unlenited forms following the same nominative plural.

_________________
The dialect I use is Cork Irish.
Ar sgáth a chéile a mhairid na daoine, lag agus láidir, uasal is íseal

Thank you for that. Airm feirmeóra is a good example. I think "na sagairt paróiste" might be acceptable, given the frequency of "sagart paróiste".