Netzoopy

-------------------------

- Added BONOBO to the zoo!
- Cobra has been updated and integrated with PANDA.

-------------------------

- Added COBRA to the zoo!
- We no longer costrain the igraph version to be older than 0.10. This will probably change the community assignment
results, but the upgrade has been recommended by the igraph developers.

-------------------------

- We have added some options to LIONESS: single lioness files can be saved in HDF (fmt='h5') which saves a lot of time
and memory. By passing ignore_final to lioness (with save_single_lioness) each lioness is discarded after being saved,
hence you won't have all lioness networks in memory at the same time.
- PANDA can be run with_header from CLI
- Added pytables/tables in dependencies.

-------------------------

- We are changing the PANDA outputs and default flags. For now we are updating the command line
call only, behavior is kept as in 0.9.11 for the internal functions. By passing `old_compatible = False`
the final output will always have column headers and indices.
- PUMA and PANDA do not save_tmp as default.
- lioness for puma has been fixed
- Fixed PANDA data preprocessing bug

-------------------------

- Added LIONESS for DRAGON with tests
- PANDA preprocessing expression: In Panda preprocessing there was a problem with indices. Using gene2idx.get(x, 0) always give you the index 0 if x is missing fro
m gene2idx.get (like a gene in gene expression and not in motif, since gene2idx is build on top of the intersection of expression and motif). Now we use gene_names to
both create the indices for self.expression and to access with .loc[] the expression data frame self.expression_data
- New PANDA tests
- Updated LIONESS start and end parameters so that they are independent of the background. Example: One can now run panda on 100 samples
and then apply LIONESS on only the first 10.
- Added LIONESS subset parameter: passing subset parameters (a list of indices or sample names, [1,2,10]) allows to run
LIONESS only on specific samples. This parameter has priority over the start and end parameters.

------------------

- Fixing single/double precision for GPU
- Clearing GPU after computation to free more memory