refactor star file reading

[alisterburt]

Here's an attempt at refactoring star file IO into something a little more flexible, keen to hear your thoughts

The main ideas behind this implementation are that:

• main function read_star reads raw data using starfile then dispatches reading
• reading is attempted by explicit "reader_functions" sequentially

I think that overall this is more explicit than the use of conventions as we were working with before which quickly becomes an unreadable nested if/else mess where interpretation is mixed with parsing. The price is a little bit of code duplication, although some refactoring could be done

relion 3.1 style shifts are now handled inline, optics and particles dataframes are merged so that shifts can be calculated properly - the implementation is robust to the possibility of multiple optics groups (which can have different pixel sizes, happens in single particle sometimes)

One thing that's lost in this implementation is the ability to use your regex wizardry for filename matching, I'm sure it can be added back in I just didn't understand it enough to do so

[codecov-io]

Codecov Report

Merging #60 (598ff49) into develop (2527ea0) will decrease coverage by 0.88%. The diff coverage is 19.67%.

@@             Coverage Diff             @@
##           develop      #60      +/-   ##
===========================================
- Coverage    54.74%   53.86%   -0.89%     
===========================================
  Files           58       61       +3     
  Lines         1233     1268      +35     
===========================================
+ Hits           675      683       +8     
- Misses         558      585      +27     

Impacted Files	Coverage Δ
peepingtom/core/datablocks/orientationblock.py	67.27% <ø> (ø)
peepingtom/io_/read/star/utils.py	15.78% <15.78%> (ø)
peepingtom/io_/read/star/star.py	25.00% <18.18%> (+6.08%)	:arrow_up:
peepingtom/io_/read/star/reader_functions.py	33.33% <33.33%> (ø)
peepingtom/__init__.py	100.00% <0.00%> (ø)
peepingtom/entry_points/peep.py	0.00% <0.00%> (ø)
peepingtom/_version.py	100.00% <0.00%> (+100.00%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 2527ea0...598ff49. Read the comment docs.

[alisterburt]

this will close #59

should also note that tests should be added - I'll try to add the test data repo with the following tomorrow

• a dynamo table file
• a relion 3.0 star file
• a relion 3.1 star file

[alisterburt]

@brisvag just summarising our discussion a bit here:

• we agreed that star file and dynamo table handling should not be handled by the same Handler
• dispatching data already read from disk is the right way to go within the star reading module, despite the nested-ness
• maybe handling star IO more generically is useful, given the duplication we see here

my feeling is still that I like these explicit handlings of specific cases despite the presence of common logic, it gives the function single responsibility and is nice and explicit

I think this is now ready to merge!