Add symmetric tridiagonal matrices module

[Mahmood-Sinan]

This draft PR adds support for symmetric tridiagonal matrices inside specialmatrices in stdlib.
It includes implementation, tests, documentation and examples.
The PR implements the same arithmetic and matrix operations as the existing tridiagonal module.

Remaining tasks:

• Add validation for the op argument in spmv, following PR Add validation for op argument in spmv routine for special matrices #1087
  (this will be added here once that implementation is merged).

Feedback is welcome.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 68.50%. Comparing base (9591c0d) to head (fc95858).

Files with missing lines	Patch %	Lines
...pecialmatrices/example_sym_tridiagonal_dp_type.f90	0.00%	7 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1091      +/-   ##
==========================================
- Coverage   68.55%   68.50%   -0.06%     
==========================================
  Files         396      397       +1     
  Lines       12746    12753       +7     
  Branches     1376     1376              
==========================================
- Hits         8738     8736       -2     
- Misses       4008     4017       +9     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[loiseaujc]

It looks pretty nice. One little thing though (even if I overlooked it when implementing the tridiagonal_type ) : you could define a single pure subroutine to construct the symtridiagonal matrices (make it two if you include the constant diagonal cases) and then have it called by the pure and impure constructors. It's a little thing but that would reduce a bit code duplication I guess.

I'll go through the code a bit more thoroughly tomorrow but it looks pretty good to me already.

[Mahmood-Sinan]

One little thing though (even if I overlooked it when implementing the tridiagonal_type ) : you could define a single pure subroutine to construct the symtridiagonal matrices (make it two if you include the constant diagonal cases) and then have it called by the pure and impure constructors. It's a little thing but that would reduce a bit code duplication I guess.

Thanks for the suggestion! You’re right, I implemented so. Now the tridiagonal and symtridiagonal construction is now handled by two pure helper routines(one from arrays and one from constants), which is called by both the pure and impure constructors.

[loiseaujc]

I'll take a deeper look at it by the end of the week, but it looks quite good to me so far.

[loiseaujc]

It is mostly a question of personal tastes, but I prefer symtridiagonal over sym_tridiagonal. The no underscore naming is also the one used by Julia. What do you guys think @perazz, @jvdp1 and @jalvesz ?

[loiseaujc]

Modulo this very style question, I think it's pretty much all good for me !

[loiseaujc]

LGTM.