[HealthAPI] Warn when ILM steps are waiting for too long

[andreidan]

ILM routinely waits on certain conditions to be true.
eg. target index is GREEN, rollover is ready

While it's normal to wait for certain conditions for a long time (ie. 30 days for rollover to be ready),
our users could benefit from being signalled that other conditions take a long time
e.g. the fully mounted index could not be allocated, the shrunk index could not be allocated in 24 hours (note that
in these particular cases the shards_availability indicator would yield a non-healthy status if primaries/replicas
are unassigned)

The health API could signal on a subset of the wait-for-XXX steps if their respective conditions
have not been met in 24 hours (configurable)

[elasticsearchmachine]

Pinging @elastic/es-data-management (Team:Data Management)

[andreidan]

Implemented via #96092