Small fixes/enhancement to ALS readers

[lbluque]

Small fixes/enhancement include:

• ALS h5 reader returns only tomo, flat, dark (unless mult_flats flag is specified) ALS hdf reader fail data load #20
• ALS tiff datasets now work correctly ALS tiff reader fail data load #21
• ALS tiff dataset reader now has sino slice keyword ALS readers #22
• ALS h5 files are handled in a context manager Follow up on https://github.com/tomopy/tomopy/pull/190#issuecomment-217037456 #24 as suggested by @tacaswell.

[tacaswell]

I do not know this code, but that sometime you are using (sino, None) and other time (None, sino) is a really bad code smell.

[decarlof]

@tacaswell it does smell bad ... thank you for point it out. For some reason we are flipping the order between datasets consisting of multiple files (tiff, xrm) vs single files (hdf, netcdf etc.). We should make it consistent.

[decarlof]

@dgursoy fix seems easy, just swap the loop indexing at

def _shape_after_slice(shape, slc):

but will make it not back compatible. Suggestions?

[decarlof]

@lbluque passed the reconstruction tests on both ALS tiff and hdf5. @tacaswell do you have a suggestion on how to run automatic testing with large data sets?

[tacaswell]

To generate synthetic data if possible, use minimal files if possible (ex a 10x10x10 tomographic dataset).

In this case, it might require a writer for each of these layouts.

[decarlof]

@lbluque I updated the docs to include two examples for reading and reconstructing the ALS data. This are at ALS examples with functions showing at API

Feel free to add/modify as needed

[decarlof]

@lbluque I am merging and closing this pull request. If you want to add more info in the documentation at ALS examples please open a new pull request.