Hello,

I am sending a new PR to solve a 'log' bug in the gromov_wasserstein2 function. Indeed, because of a variable name mismatch, the log was overwritten and was always returned.

In this PR, you will find:

Bug 'log' correction
Test log=False for gromov_wasserstein2
Removing 'H entropy' from doc for gromov_wasserstein and gromov_wasserstein2

numItermax bug for emd function inside cg function. The number of iter was fixed to 100.000 and could not be modified.
Test numItermaxEmd > 100000 for cg.