[victorapm]

In summary, this PR:

• Initialize open wells before time-zero output so pressure and stored fluid are available.
• Honor initial well and perforation schedules and validate phase-rate constraints.
• Preserve deferred initialization for initially closed wells.
• Add open/closed initialization regressions; compositional, single-phase, thermal, and scheduling tests pass.

Additional details

Initially open wells were previously initialized during the first coupled solver step. At time zero, their pressure and component-density fields could therefore remain zero. This also omitted fluid already stored in the wells from the initial reservoir-plus-well inventory, making subsequent mass-balance comparisons misleading.

This patch initializes open wells after the reservoir fluid state and well separator models are available, using the existing well-initialization path. It evaluates the well and perforation schedules at time zero and validates the active well constraints before initialization. That validation establishes the phase indices needed by phase-rate controls.

Wells that are closed initially retain their deferred initialization until opening. New regressions verify that open wells have populated pressure and component densities before any solver step, while an initially closed well remains uninitialized.

Validation passed: testOpenClosePerf, testReservoirCompositionalMultiphaseMSWells, testReservoirSinglePhaseMSWells, and testThermalReservoirCompositionalMultiphaseMSWells.

[tjb-ltk]

There needs to be a test showing the problem. Initialization of the wellbore during first opening can be done a few ways, including a well estimator. Its not clear what is specifically being addressed in this PR