Update prox-grad backtracking condition

[hajg-ijk]

The :convex backtracking strategy for proximal_gradient_method now entails a slightly different condition whenever the upper bound on the sectional curvature of the manifold, input via the k_max kwarg, is positive. This comes with a "tolerance" type parameter, δ, which must be positive.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (0e92b74) to head (2868f2e).

Additional details and impacted files

@@            Coverage Diff            @@
##            master      #606   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           91        91           
  Lines        10052     10054    +2     
=========================================
+ Hits         10052     10054    +2     

☔ 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.

[kellertuer]

Is this breaking existing behaviour? Then it should probably be part of the forthcoming 0.6 release?

Besides that one of your stopping criteria is now not code cov'ed on 2 lines, StopWhenGradientMappingNormLess never reaches the point where it reports true.

[hajg-ijk]

Is this breaking existing behaviour? Then it should probably be part of the forthcoming 0.6 release?

Besides that one of your stopping criteria is now not code cov'ed on 2 lines, StopWhenGradientMappingNormLess never reaches the point where it reports true.

I don't think this is breaking. But I am a bit puzzled by those 2 missing lines. Will try to cover them.