Replace USE_COEFFICIENTS with templates

[reds-heig]

Hello !

This PR remove USE_COEFFICIENTS define in the C++ code and replace all the related code with templates.
Now, when running Ripser with modulus==2 the computation should be slightly faster. Following tables shows some datasets run with coeff==2:

Dataset	size	threshold	dim	coeff	with USE_COEFF [s]	templated [s]
sphere3	192		2	2	1.4	1.1
dragon	2000		1	2	2.6	2.2
o3	1024	1.8	3	2	2.3	2.0
random16	50		7	2	6.2	5.9
fractal	512		2	2	15	14
o3	4096	1.4	3	2	53.9	50

About the changes done inside the C++, I'm currently not happy with my solution of removing because of:

• simplex_coboundary_enumerator isn't anymore a nested class but a friend class
• get_edges isn't anymore a method of ripser class but a friend function
• get_vertex_birth isn't anymore a method of ripser class but a friend function

I needed to remove simplex_coboundary_enumerator from a nested class, because I could not find a solution to partial specialization, but please feel free if you have a better alternative !
Same remark for get_edges and get_vertex_birth

Additional stuff I could add:

• Update Changelog
• Refactor C++ backend with .clang-format of ripser (I found the file)
• Documentation ??

Best,
Julián

[codecov]

Codecov Report

Merging #116 (452fda3) into master (f784e1f) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #116   +/-   ##
=======================================
  Coverage   96.75%   96.75%           
=======================================
  Files           3        3           
  Lines         154      154           
  Branches       26       26           
=======================================
  Hits          149      149           
  Misses          4        4           
  Partials        1        1           

---

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 f784e1f...0b267eb. Read the comment docs.

[MonkeyBreaker]

One last point, please feel free to validate from your side, I tried to run with different modulus variables and so far I got expected results. But if on your side you have some use cases with modulus>2, validate it on your side, so I can be more confident I didn't break anything :)