CAMEL-24512: camel-spring-boot - keep each failing health check's error message

[oscerd]

CamelHealthCheckIndicator.doHealthCheck loops over every health check result and passes the same
Health.Builder to CamelHealthHelper.applyHealthDetail. The error message was written only as a flat
top-level detail:

builder.withDetail("error.message", error.getMessage());

Health.Builder.withDetail is a Map.put, so that key is a single slot shared by every check. With two DOWN
checks the actuator response carried only the second message:

error.message=second-message
first.data={...}
second.data={...}

The per-check <id>.data maps stay separate, so the fix is to put the message there too. The top-level key is
kept as it was, for compatibility.

Found by @luigidemasi while reviewing CAMEL-24499; it pre-dates that change and is independent of it.

Scope

Deliberately confined to error.message. #1929 (CAMEL-24592) changes the stack-trace gating in the same
method, and @Croway scoped that PR to leave this alone — doing the same in reverse here, so the two should
apply cleanly in either order.

Tests

CamelHealthHelperMultipleFailuresTest:

• eachFailingCheckKeepsItsOwnMessage — two DOWN results with distinct messages, asserting each is retrievable
  under its own check id
• topLevelErrorMessageIsStillReported — the compatibility guard on the flat key

Verified meaningful: the first fails against main (expected <first-message> but was <null>), the second
passes either way. Full core/camel-spring-boot suite: 163 tests, 0 failures.

Note on the root build

The full reactor currently fails on camel-hivemq-starter, unrelated to this change:
org.apache.camel:camel-hivemq:4.23.0-SNAPSHOT is not on apache.snapshots yet (main's HEAD is the commit that
added the starter). 171 modules build, core/camel-spring-boot among them, before the reactor stops there.
Same publishing-lag shape as camel-toon a few days ago.

🤖 Generated with Claude Code

https://claude.ai/code/session_0165HC1XCB3h6mMii6wbZneG