Remove objective-direction compatibility shims and enforce canonical energy semantics

[Copilot]

This PR completes the objective-direction migration as a clean break: deprecated compatibility APIs and alias fields are removed, and the codebase now exposes a single canonical invariant based on objective_sense, objective_value(...), and energy(...). QAOA internals remain energy-first and argmin-based throughout optimization, statistics, warm-starting, post-processing, and serialization.

• API cleanup

  • Remove deprecated compatibility methods:
    • Problem.cost()
    • Problem.computeMinMaxCosts()
    • QAOA.get_Exp()
  • Remove legacy result/landscape alias fields that duplicated canonical energy-named fields.
  • Update tests and docs to use the canonical API only.

• Canonical objective/energy semantics

  • Keep one invariant across the codebase:
    • energy(x) = objective_value(x) for minimize problems
    • energy(x) = -objective_value(x) for maximize problems
  • Remove fallback/adapter logic that preserved old sign conventions.
  • Keep warm-start selection, CVaR/statistics, best-solution tracking, and post-processing energy-native.

• QAOA internals

  • Rename internal grid-search/evaluation paths to energy terminology.
  • Keep landscape sampling and depth expansion strictly based on minimizing energy.
  • Remove compatibility aliases from post-processed and sampled results.

• Serialization / schema

  • Keep serialized payloads on canonical objective/energy semantics only.
  • Bump qaoaIO schema to 3.
  • Reject legacy payloads instead of silently adapting old compatibility shapes.

• Docs / examples / migration surface

  • Remove backward-compatibility language from the README.
  • Document the clean-break API explicitly:
    • objective_value(...)
    • energy(...)
    • objective_bounds()
    • get_energy()
    • get_objective()
  • Update example notebooks and fixture data to the canonical API.

Example of the resulting API:

value = problem.objective_value(bitstring)
energy = problem.energy(bitstring)

best_energy = qaoa.get_energy(depth=1)
best_objective = qaoa.get_objective(depth=1)

[fgfuchs]

this is jsut to updat ethe branch