I very much appreciate these questions. I am in favor of a very strict style; or in other words, helping students to avoid errors and confusions in the standard language as much and as early as possible. The main thing that sticks in my mind is that the longer a bad practice persists, the "stickier" it is, and the harder it is to fix later on.
Comparing (a) learning a thing right the first time, vs. (b) learning a thing wrong and fixing it later, item (b) takes roughly three times as much effort as (a). That's because (b) entails (1) learning the thing wrong initially, (2) working to lose the erroneous idea/muscle memory, and (3) finally learning the thing right.
Anecdote: I was teaching a remedial algebra class and reviewing the arithmetic order of operations. As part of the discussion, it became clear to me and a young woman of color that she had flat-out been taught something totally incorrect at the high school level. She burst into tears and fled from the classroom, enraged that, "she'd been taught everything wrong and had to learn it all over again".
Now, that was an outlier reaction, but it's always stuck with me. She's right, and I respect her high level of intellectual honesty. Not bothering to get these fundamentals right early on is something like a brutality to the students in question. Not caring about it is an equity issue.
Let's compare to an analogous trend a few decades ago in English instruction; to get rid of directed instruction in phonics, in favor of a more flexible, exploratory, "whole word" learning of vocabulary. In short, the results were that such a removal of direct instruction was clearly detrimental. From a meta-study reported by Brady, Susan A. "Efficacy of phonics teaching for reading outcomes." Explaining individual differences in reading: Theory and evidence (2011): 69-96:
Overall, research reviewed in the NRP report indicates that students taught with systematic phonics instruction have better reading scores, whether measured at the end of the training period or at the end of the school year of instruction (Cohen's d = .44), Systematic phonics instruction was found to produce better reading growth than all of the types of nonsystematic or nonphonics instruction (i.e., basal programs, whole-language approaches, regular curriculum, whole word curriculum, and miscellaneous programs). Further, systematic phonics was found to be effective whether taught through individual tutoring (d = .57), through small groups (d = .43), or to the whole class (d = .39).
My impression is that likewise, the elimination of directed grammar instruction and correction is now producing evidence of serious corrosive effects; but I'm not up on that research and would need to go digging for specifics.
In summary, all the signs I can gather -- both from in-class experiences and decisive evidence from language-instruction research -- point toward it being better to get the fundamentals right sooner rather than later.