[BUG] LSTM model does not have encoder_activation parameter but contains in the docstring

[yarnabrina]

I've noticed that LSTM model does not have encoder_activation as an argument in __init__ but it's documented in the docstring.

neuralforecast/neuralforecast/models/lstm.py

Line 31 in 7f69d78

`encoder_activation`: str=`tanh`, type of LSTM activation from `tanh` or `relu`.<br>

[jmoralez]

I think this is the same as #426

[yarnabrina]

Yes. If you want, this can be closed as duplicate.

Will you accept a PR removing the line from docstring? The alternative by adding a new argument (similar to how it's there for RNN) will likely be a breaking change for users relying on order of keyword arguments.

[jmoralez]

Will you accept a PR removing the line from docstring?

Yes, it doesn't seem to be used.