MeldaAutoAlign, sub-sample precision, dynamic time-alignment correction

[Nspace]

I came into Precision Time Align by Eventide. Its description is similar to other generic time aligners, except it declares itself having "High precision, sub-sample delay. Get perfectly synchronized double-tracked recordings and significantly eliminate timing anomalies and phase issues from your mix on multi-mic’d drums, vocals, and instruments. Precision Time Align allows for synchronizing signals to microsecond accuracy, up to 1/64th of a sample.".

And having myself worked (for more than a decade!) looking down-to sample sized timelines, it appears to me as quite relevant to get summing files into the closest precision alignment attainable, more so for files with relevant high frequency content...
So what is the state of things with MeldaAutoAlign about sub-sample precision?

This question put forward, I'd also like to understand the rest of the audio alignment available features in the plugin world, compared with how actually "MAutoAlign performs something previously unimaginable - it actually fixes the phase for each frequency in all tracks, if you ask it to. During the analysis it now not only estimates the ideal delays and polarity, but also the ideal phase rotation to minimize cancellations." As it Melda webpage describes, specifically knowing what it does? Does it cleverly delay the start time position of each track or... in fact it applies inside-of-the-file variously timed delays where appropriate or some other algorithm based alterations to the audio?

SoundRadix developers release two plugins differentiating these two features; Auto-Align 2 (seems to be similar to MAutoAlign), in their words "Absolute Phase Optimization in Auto-Align 2 goes beyond aligning the microphones in phase with each other. It also corrects the overall sound directionality to make sure the reproduced sound matches the original source. This means that sounds that were originally pushed forward will also be reproduced forward in the speakers.".
And then, Auto-Align Post 2 which additionally "Enabl(es) dynamic time-alignment correction of moving location microphones, Auto-Align Post makes it possible to effectively mix the boom and lav microphones by eliminating comb-filtering and phase artifacts caused by the time-of-arrival differences, with just a few clicks.
The new Spectral Phase Correction module detects and dynamically corrects frequency-dependent phase shifts to achieve significantly improved spectral alignment and richer sound".

So Auto-Align Post 2 has two functioning modes: "Dynamic mode enables continuous phase/time correction for moving actors or cameras. Static mode enables fixed phase/time correction for stationary microphones". And version 2.1 adds a Spectral Correlation Meter. "A spectral magnifying looking-glass provides a detailed 28-bands frequency phase correlation view.".

//All in all, the above looks as quite interesting functions and features. How MAutoAlign compares in terms of features, precisely?

[MeldaProduction]

I'm not sure of course, but I believe this SR plugin is probably doing what we have for a long time there as the spectral phase correction. Note that while this sometimes does magic, other times it only garbles the sound. Unfortunately that's kind of a physics thingy, you can't cheat time .

Anyways as for microdelay - to me it seems pointless. If you think about it, a delay lower than one sample would only produce a comb filter slightly below nyquist, so it's effect would be more like a very very slight lowpass, similar to interpolation which they would have to use. So imho it's a waste of CPU power, which just produces slightly different artifacts.