fix(naming): Recreate entity IDs proposes what the panel says now

[cayossarian]

Closes #252.

Renaming a circuit in the SPAN app left "Recreate entity IDs" offering the entity
its own ID, so the button appeared to do nothing.

Cause

The registry generates an ID from the user's name override, else
suggested_object_id, else object_id_base. Friendly-names mode writes no
registry name — the panel name arrives as original_name — so the suggestion
decides. construct_single_circuit_entity_id was handing back the entity's
stored ID on every reload, which HA recorded as the suggestion, so the field was
frozen at whatever the circuit was called the day it was added.

Fix

It now computes the ID from current panel data and the naming flags, for existing
entities as much as new ones.

The three constraints, and why each holds

• Existing entity IDs cannot move. An entity_id set before add is only a
  suggestion: EntityPlatform splits it into suggested_object_id, and for a
  unique_id already on file async_get_or_create routes to
  _async_update_entity with no new_entity_id, then reassigns
  entity.entity_id from the stored entry. Structural, not a matter of care.
• Unique IDs cannot move. They derive from the description key and never read
  a name. Untouched here.
• Phase 2 name sync works as before. Untouched. All six registry-name write
  sites are gated on USE_CIRCUIT_NUMBERS, and nothing in that mode changes.

Circuit-numbers mode is deliberately unchanged

There the registry name phase 2 sync writes is both what shows the panel's name
in the UI and what outranks the suggestion during regeneration — one write, two
effects. So Recreate in that mode keeps composing from the name rather than the
tab-based ID. That is pre-existing, and a test pins it so this change cannot
disturb it. Correcting it means rerouting phase 2 sync, which is a product
decision recorded in the design doc, not part of this fix.

Cleanup

The registry lookup was the only use of the unique_id argument, and
existing_entity_id the only use of that parameter on _construct_entity_id.
Both are removed rather than left as parameters that no longer decide anything.

Tests

Eight new cases, all reloading before asserting — asserting straight after
creation exercises the first-add path, where the suggestion is trivially current
and the bug cannot appear. Three demonstrate the bug and fail on the previous
code (sensor suggestion, async_regenerate_entity_id, breaker switch); five are
guards that pass on both (entity ID unmoved, unique ID unmoved, unrenamed circuit
offered its own ID, circuit-numbers mode unchanged).

1261 passing, up from the 1253 baseline. Full prek hook set green, mypy strict
and vulture included.