chore(presentation): consolidate canRequestRide predicate into FollowedDriversRepository.canRequestRide(_:)

[variablefate]

Background

DriverDetailViewState.from() and RideRequestDriverOption.from() inline the
predicate driver.hasKey && !isKeyStale && location?.status == "online" for
canRequestRide, instead of calling the SDK-side
FollowedDriversRepository.canRequestRide(_:) that already exists.

History

This drift was first surfaced in PR #58's 4th-pass review and re-flagged in PR
#61's review as a follow-up to fix at next touch. PR #93 (Kind 30173 vehicle
work) did modify DriverDetailViewState.swift (added a vehicle: parameter)
but kept the inline predicate to honour PR #91's "don't widen scope" constraint.

Why it matters

• The inline predicate omits the "driver exists in repo" guard pinned by
  CanRequestRideTests.staleCallerSnapshot_hasKeyButRepoMissingKey_returnsFalse
  on the SDK side. Practically harmless today (callers always pass a real
  FollowedDriver), but a real semantic gap.
• Two divergent definitions of "rider can request a ride from this driver" make
  future protocol-level changes (e.g. new key staleness conditions) risky:
  callers updating one definition and forgetting the other would silently drift
  the UI from the SDK contract.

Scope

Consolidate the inline predicate in:

• RoadFlare/RoadFlareCore/Presentation/DriverDetailViewState.swift
• RoadFlare/RoadFlareCore/Presentation/RideRequestDriverOption.swift

into calls to FollowedDriversRepository.canRequestRide(_:). Update the
factory signatures to accept canRequestRide: Bool (resolved by
AppState+Presentation) rather than re-deriving from raw inputs. Update
existing tests to feed the resolved boolean.

Effort

Small — one-shot refactor + test churn. No protocol changes.