Make Egret aware of APPSI

[bknueven]

We have some special logic in the unit commitment and DC/SC OPF methods for Persistent solvers. We should make this same code aware of APPSI solvers. This would especially improve performance for CBC.

[bknueven]

Now I remember the hold-up on this: @michaelbynum how amendable would you be to adding a add_column method for the APPSI solvers similar to that for PersistentSolver? Here's an example of how lazy transmission constraints are currently handled:

Egret/egret/common/lazy_ptdf_utils.py

Lines 621 to 640 in 3ddc971

	constr[bn], new_slacks = _generate_branch_thermal_bounds(mb, bn, thermal_limit)
	viol_in_mb.append(i)
	if new_slacks:
	m = model
	obj_coef = m.TimePeriodLengthHours*m.BranchLimitPenalty[bn]
	if persistent_solver:
	m_model = m.model()
	if m is not m_model and obj_multi is None:
	raise RuntimeError("Cannot add lazy var for branch slacks if part of a larger model")
	if obj_multi is not None:
	obj_coef = obj_multi*obj_coef
	## update the objective through the add_column method
	solver.add_column(m_model, mb.pf_slack_pos[bn], obj_coef, [], [])
	solver.add_column(m_model, mb.pf_slack_neg[bn], obj_coef, [], [])
	else:
	m.BranchViolationCost[time].expr += ( obj_coef*mb.pf_slack_pos[bn] + \
	obj_coef*mb.pf_slack_neg[bn] )
	if persistent_solver:
	solver.add_constraint(constr[bn])

[michaelbynum]

Ahh, yes. Definitely. Will do.