ACSTP acstp

included normalization and smoothing of the SNR estimation for the subspace dimension selection (P_z). More robust.

Error corrected in verification INPUTS LatencyCorr_max and Epoch_size. This error was happening when the user did convert artificially the epoched data to continuous EEG.

Needed to normalize the SNR in best_Pz.

the ACSTP beta works with pop_acstp under the eeglab structure with few limitations. First of all, not all the parameters needed are proposed in the pop_acstp.m.
I reported all the observation under comments with the tag "%*LK:"
Moreover, I added some warning for debugging.

script_test_pop_acstp.m has 2 error examples as well as a working example.

The output of the working example is placed in the repertory .\pic\

Resubmit to try to break the error from SourceForge

Added (again) eegplugin_acstp.m and pop_acstp.m

New version with plenty of new features in ACSTP.m
Now the variables have more straightforward names and I included a lot of Input verifications.