Allow lbfgs on auglag, expose more kwargs and fix max_eval

[mohamed82008]

This PR does 3 things. Let me know if you prefer I split them.

• By introducing the modifier kwarg, one can now pass a function that changes the augmented Lagrangian model to an LBFGS model
• I exposed more kwargs from tron to percival to give more options to set.
• I made sure max_eval referred to the total number of function evaluations not for each loop iteration. Let me know if you think a different kwarg would be better here.

[codecov]

Codecov Report

Merging #44 (6fe9d08) into master (2132510) will increase coverage by 0.20%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #44      +/-   ##
==========================================
+ Coverage   90.20%   90.41%   +0.20%     
==========================================
  Files           3        3              
  Lines         143      146       +3     
==========================================
+ Hits          129      132       +3     
  Misses         14       14              

Impacted Files	Coverage Δ
src/method.jl	84.26% <100.00%> (+0.54%)	⬆️

---

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 2132510...6fe9d08. Read the comment docs.

[abelsiqueira]

Thanks for the PR. I suggested changes for the names:

• modifier -> subproblem_modifier
• tron_max_eval -> subsolver_max_eval
  And I'm thinking that maybe max_cgiter could be passed as a Dict subsolver_kwargs because its splats nicely and when/if we add new subsolvers, it won't need much change.

[abelsiqueira]

Let me know when it's time for a new review with the "request re-review" button, please.

[abelsiqueira]

Thanks