Add configurable inequality tolerance to the VMCON solver

[timothy-nunn]

Modifies the default behaviour changed recently in #3858. Instead of applying no additional convergence criteria, we will revert back to applying the old inequality convergence criteria but with a default relative tolerance of 1e-8 (was 0 before) that can be modified in the IN.DAT. The option to turn this additional convergence criteria off will remain (introduced in #3858).

[codecov-commenter]

Codecov Report

❌ Patch coverage is 80.00000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 46.44%. Comparing base (124a7ee) to head (e14288a).

Files with missing lines	Patch %	Lines
process/solver.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #3878   +/-   ##
=======================================
  Coverage   46.44%   46.44%           
=======================================
  Files         123      123           
  Lines       28044    28047    +3     
=======================================
+ Hits        13024    13027    +3     
  Misses      15020    15020           

☔ 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.

[jonmaddock]

Great, I think this is a good compromise: not completely strict (convergence trouble) but not the original VMCON convergence criterion either (allowed large inequality violations if Lagrange multiplier was small). This gives us the original VMCON convergence criterion whilst additionally ensuring we don't have large inequality violations.