I don't know very much about Vedic meter, but I think in general we must be aware that we are not only dealing with a fairly long span of time but also with changes in genre. The later point is very important for Sanskrit during the classical period because texts continued to be written in different genres that have different standards for grammar and meter. The rules for Kāvya are fairly strict and that is why commentators tend to discuss, defend, reject, and at times accept (even if begrudgingly) various anomalies. A few of these debates I have tried to follow somewhat exhaustively (or rather ad nauseum) over the course of 1400 years and in as many sources as I could access. I would emphasize two things: there is usually great variety in how commentators understand certain forms, which grammars/commentaries they hold to be authoritative, and how they interpret Pāṇini's (or Candra's etc.) rules; 2. we have access to a fairly limited amount of the discourse, especially if we think of how many millions of people read and discussed mahākāvyas like the Raghuvaṃśa, so one should be cautious with the interpretation of the data and generalizations.
When people wrote in very obviously non-standard Sanskrit, they generally did so under the name of a great sage or with the claim of divine revelation, etc. This gives the author more leeway to include archaisms and other non-standard forms. The Bhāgavatapurāṇa is perhaps the greatest example.
Specifically on meter, I don't know what the communis opinio is anymore, but I recall that with the supernumerary syllables in Vedic meters---and with other oddities---we should perhaps not be so quick to restore or label such things as "wrong". In later times, the Vedas simply are and are not subject to any rule or criticism. I don't believe we have any roughly contemporaneous criticism of Vedic poets so it is difficult to know how such meters were perceived at or around the time of their composition. For classical kāvya-meter, I know of only one real issue that arrises from time to time and that is the application of the muta cum liquida rule, i.e, that a short vowel before consonant clusters like pr are counted as laghu. I have spoken with some other colleagues who have noticed that it seems to apply more frequently in Purāṇic-like texts and hope to write something about, but perhaps an article already exists on the topic. In case this should be of interest to anyone, I simply paste below the email I sent to my students some time back.
All the Best,
Victor
Just a short note about meter. Thanks to a helpful conversation with Vladimir the other day, it has come to my attention [ultimately from Haru] that I should, for the sake of completeness, mention that the basic rules for determining a guru syllable require an addendum. In the overwhelming majority of cases, a short vowel followed by any two consonants will be counted as metrically heavy (guru). However, there are a few rare occurrences where we can apply the so-called muta cum liquida rule, which might be familiar to some of you from Latin and Greek metrics. In short, a stop plus l or r need not make the preceding syllable heavy, i.e., it can be either laghu or guru. For example,
albānīque patrēs, atque altae moenia Rōmae.
"And the Albanian fathers as well as the walls of lofty Rome"
Here pa is scanned as laghu before tr and the second foot is a dactyl (= bha-gaṇa).
In Sanskrit such an indulgence is rather rare and appears to be more frequently applied in Prakrit poetry and "Epic" Sanskrit. In the latter case several types of consonant clusters do not cause the preceding syllable to be guru. See Oberlies
A Grammar of Epic Sanskrit p.
XXXVII. For Prakrit, see Prākṛtapiṅgalasūtra 1.4 (
https://archive.org/details/KavyamalaVol41PrakritaPingalasutraniOfLakshmanabhatta1894_201803) and the examples in the commentary by Lakṣmīnātha Bhaṭṭa:
ceja sahaja tuhũ cañcalā sundarihradahĩ valanta /
paaü ṇa ghallasi khallaṇā kīlasi uṇa uhlasanta //
"My heart! you are by nature fickle [yet] you've returned to the abyss of this beauty. You do not set foot [anywhere else]. You fool! yet again you play [there] trembling with delight."
The point beging that for the Doha meter (13-11-13-11 mātrās) to work, the i before hr and the u before hl must be scanned laghu.
For Sanskrit, see Vṛttarantākara 1.10f. with the commentary of Bhaṭṭanārāyaṇa. The kārikā itself is rather restrictive (only before the beginning of a pāda), but in the commentary, Bhaṭṭanārāyaṇa furnishes additional examples from Kālidāsa (Kumāsambhava 7.11b gṛhītapratyudgamanīyavastrā) and Māgha (Śiśupālavadha 10.50a prāpya nābhihradamajjanam āśu), as well as from the Rāmāyaṇa. In the examples, which contain variant readings from standard editions, the a before pra and the i before hra must be scanned laghu.
There are further discussions (Sarasvatīkaṇṭhābharaṇa 1.123 and the footnotes to Piṅgala's Chandaḥśāstra pp. 4f., ed. Kedāranātha 1938), and Śaraṇadeva also mentions the problem as well as the variant reading of Śiśupālavadha 10.60 with hrada in his Durghaṭavṛtti ad P. 1.4.11, but alaṃ vistareṇa.