Re-add verbose argument to `load_xdf` about pyxdf

[chkothe]

I think the strongest argument is probably to not break backwards compatibility (even though pyxdf is relatively new). But other than that, and acknowledging that the verbose flag was probably carried over from some MATLAB code, it's more convenient to pass a bool into a verbose flag than to work out and manicure some configurable per-module logging setup, and pass the bool into that.

I think this package is mostly useful for data scientists and neuroscientists, who I think appreciate a quick solution that gets the job done without much software engineering -- or to not necessarily technical end users of larger packages in the style of EEGLAB or MNE, where a lot of the details and responsible modules are sometimes obscure to them, but one can at least hope that all the options of the underlying import functions are exposed somehow by that kind of software.

Unfortunately there's only a single "debug" log level in Python, and pyxdf is rather spammy at that level, making application-wide debug logs pretty hard to traverse out of the box without such a facility in place.

[cboulay]

Are you advocating for not using the logger at all? Or only when verbose=True?

[chkothe]

IIRC it was just suppressing some of logger messages when set to False (probably all, but that's maybe a bit too much). We could make it suppress just the detailed per-chunk messages if False.

[cboulay]

Can we default to verbose=True? People who rely on logger entirely might want to ignore the verbose argument yet still get the debug statements (because that's how they configured their logger). If so then the verbose argument would effectively be equivalent to allow_debug_logging, though we won't name it that.

Relevant link for logging levels

[chkothe]

I think that would make sense.