Pinning fixed versions for dependencies

[m-a-r-i-u-s]

Fixed versions for the dependencies could lead to some troubles for the users as mentioned in this conversation .

TODO

Check possible ranges for the following packages:

• hmmlearn
• networkx
• numpy
• scikit-learn
• script

[m-a-r-i-u-s]

With this entries the unit tests passes
"MDAnalysis>=2.0.0",
"hmmlearn == 0.3.2",
"networkx >= 3.0.0",
"numpy >= 1.26.0, <2.0.0",
"scikit-learn >= 1.4.0",
"scipy >= 1.13.0"

I am not entirely sure about NumPy 2.0 yet. What do you think @yusuferentunc ?

-> Using NumPy 2.0.1 led to an error in the unit tests: ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject`

[yusuferentunc]

I think NumPy > 2.0.0 versions had some incompatibility with scikit-learn version or scipy version. Besides that, I think it looks good.

Current problem is hmmlearn and networkx has some problem with pypi package installation. Somehow, you can install it via pip install hmmlearn networkx command but when encapsulating into one package pip install domhmm command throws error about can't find packages hmmlearn and networkx. I was working on this problem.