Bad behavior if letter case of eventtype name is incorrect (in cfg.eventypes)

[olafdimigen]

uf_designmat looks for 'eventstypes' but using strcmpi(), which ignores case.

Thus, when an event is specified with a wrong case, e.g.
cfgDesign.eventtypes = {'sTiMuLuSoNsEt'} % should be 'stimulusOnset'

then the modeling process does start (no error produced initially), but mayhem happens later at the uf_glmfit stage (without producing an explicit error)

[olafdimigen]

Changed to be case-sensitive in commit 1a38d40

[behinger]

this is breaking change, so in principle we should bump Unfold Version to 2.0

[olafdimigen]

Tested this a bit more deeply now.

I agree that this is potentially breaking, but it looks like it can't break anything that is not already severly broken.
If an existing user did use an incorrect letter case for any event in the model, they either...

(1) got an immediate error (but this only happens if there is only this one predictor-event in the model!), or
(2) got incorrect betas (i.e. zeros for the predictors associated with this event) or,
(3) uf_glmfit() seemingly runs forever

Finally, I also saw weird warnings by the solver, but unfortunately cannot reconstruct that scenario (may be the same as the "runs forever" one).

So the downstream consequences of case-insensitivy are bad. For example, the non-time-expanded DM will still look okay, initially (with entries in all columns) but only before uf_epoch is run, which removes the epochs for the misspelled events. For Xdc, colums will be zeroed for the misspelled event.