Fix failsafe composition and update README

[rameshmalla]

Fixes #1859.

Description

Merges the four separate per-policy FailsafePlugin instances (retry, circuit-breaker,
backup-request, timeout) that riptide-spring-boot-autoconfigure registered per client into a
single FailsafePlugin per client, sharing one Policy list and one executor.

Key changes:

• FailsafePluginFactory: replaced createRetryFailsafePlugin / createCircuitBreakerPlugin /
  createBackupRequestPlugin / createTimeoutPlugin with one createFailsafePlugin(...) that
  builds a combined List<RequestPolicy> (timeout → backup-request → retry → circuit-breaker)
  and applies it via a new FailsafePlugin.withPolicies(List<RequestPolicy>) method
  (riptide-failsafe).
• DefaultRiptideRegistrar: replaced the four separate
  register*FailsafePlugin(...) methods with registerFailsafePluginWithConfiguredPolicies(...),
  which registers one FailsafePlugin bean per client whenever any of
  retry/circuit-breaker/backup-request/timeouts is enabled, and wires the (now optional) circuit
  breaker bean into it directly.
• Introduced a single, unified thread pool config riptide.clients.<id>.failsafe.threads, shared
  by all four policies on a client. The old per-policy retry.threads /
  circuit-breaker.threads / backup-request.threads / timeouts.threads settings are no
  longer supported — configuring any of them now fails fast at startup with a new
  LegacyFailsafeThreadsException pointing at the replacement property.
• Updated RiptideProperties/Defaulting to add the new Failsafe/threads config, and updated
  tests (FailSafeExecutorAutoConfigurationTest, PluginTest) and the
  riptide-spring-boot-autoconfigure README accordingly.

Motivation and Context

Having four independent Failsafe instances per client (one per policy) meant their internal
guards/executors didn't coordinate with each other, which could lead to deadlocks under
concurrent retry/circuit-breaker/backup-request/timeout execution on the same request. Merging
them into a single FailsafePlugin with one shared executor fixes that coordination problem.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• Breaking change (fix or feature that would cause existing functionality to change)
  — legacy per-policy *.threads settings now throw an exception at startup instead of being silently honored.

Checklist:

• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have added tests to cover my changes.

[fatroom]

👍

[rameshmalla]

👍