Feat: Class-based solvers

[mrava87]

This PR creates class-based solvers following PyLops' pylops.optimization.basesolver.Solver base class.

As part of this effort, all function-based solvers are modified to have the same set of tolerances exposed to the users:

• tol: tolerance on objective function change between iterations (as (cost_previous - cost_current ) / cost_previous < too directly handled in the step method
• rtol: relative tolerance on objective function vs initial value (handled as callback)
• xtol: absolute tolerance on change of solution - currently parked as requires a new PyLops release that fixes https://github.com/PyLops/pylops/pull/762/changes. This will simply require adding

  if xtol is not None:
       callbacks.append(ModuleUpdateCallback(xtol))
  

  to each function based solver.

Note that a breaking change is introduced in PPXA and ConsensusADMM as xtol was used to check a change in the solution, whilst now it becomes aligned with all other solvers for cost function update. Similarly, xtol in the AdaptivePrimalDual was previously used for x/y updates and is not used for cost function update (whilst a new tolerance xytol is added for x/y update).

[codacy-production]

Not up to standards ⛔

🔴 Issues 4 high · 7 medium

Alerts:
⚠ 11 issues (≤ 0 issues of at least minor severity)

Results:
11 new issues

Category	Results
ErrorProne	4 high
Complexity	7 medium

View in Codacy

🟢 Metrics -28 complexity · 34 duplication

Metric	Results
Complexity	-28
Duplication	34

View in Codacy

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}