replace inf bounds before setting param.maxUB to avoid unbounded solution

[PkiwiBird]

replace inf bounds before setting param.maxUB to avoid unbounded solution

I hereby confirm that I have:

• [ x] Tested my code on my own machine
• [ x] Followed the guidelines in the Contributing Guide
• [x ] Selected develop as a target branch (top left drop-down menu)

(Note: You may replace [ ] with [X] to check the box)

[github-actions]

Summary:

Tests 📝	Passed ✅	Failed ❌	Skipped ⏭️
253	218	0	35

Failed Tests:

No failed tests ✨

Github Test Reporter by CTRF 💚

[rmtfleming]

infeasibility \ne unbounded

[PkiwiBird]

This PR is correcting a mistake I made in the previous PR that was already merged. I originally wrote:

if ~isfield(param,'maxUB')
   param.maxUB = max(max(model.ub),-min(model.lb));
   param.minLB = min(-max(model.ub),min(model.lb));
end

before:

infLB = (model.lb == -inf);
infUB = (model.ub ==  inf);

%add a large finite lower bound here
model.lb(infLB) = -1/feasTol;
%add a large finite upper bound here
model.ub(infUB) = 1/feasTol;

Since the lifted gecko model had some inf in ub, param.maxUB became Inf then propagated into relaxFBA_cappedL1 as the cap for the relaxation slacks (p, q, r), so the slacks become unbounded. that's why now I changed the order and put the 2nd block first.

The PR also now corrects another issue: models without D were failing at rmfield(model, {'ctrs','dsense','d','C','D'}). Now C/D are only removed when they exist

[github-actions]

Summary:

Tests 📝	Passed ✅	Failed ❌	Skipped ⏭️
257	222	0	35

Failed Tests:

No failed tests ✨

Github Test Reporter by CTRF 💚