fix: skip field processing when nested property has same name

[MikeEdgar]

Fixes #2687

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[phillip-kruger]

LGTM. I built the branch and ran the core tests (410 run, 0 failures). I also compared the output against main for a few variations:

Scenario	main	PR
Record from the issue: record Colliding(@JsonUnwrapped Inner foo)	only bar	bar, foo ✅
Bean with private @JsonUnwrapped PInner foo and public getFoo()/setFoo()	only bar	bar, foo ✅
@JsonUnwrapped(prefix = "x") PInner xfoo (collision after the prefix)	only xbar	xbar, xfoo ✅
@JsonUnwrapped @Schema(description = "outer") PInner foo	only bar	bar, foo ✅

I first suspected the outer accessor (foo()/getFoo() returning Inner) would attach to the inner foo resolver now that it's no longer ignored. updateTypeResolvers only stores the method when the types match, though, so it's fine.

Some minor, non-blocking comments:

1. Edge case (FYI). If the outer class already has a property serialized as foo, for example @JsonProperty("foo") Integer other alongside @JsonUnwrapped PInner foo, the result is now foo: {type: [integer, string], format: int32}. main gave just integer, though it dropped the unwrapped property. Jackson would write a duplicate key here anyway, so this is a user error, but the merged schema is a bit odd. It could be handled in a follow-up, or maybe a warning.
2. Comment wording. "Since the field will be ignored" could say more about why the early return matters, e.g. "...returning here avoids marking the unwrapped property as ignored."
3. Javadoc placement in the test. The /** Issue ... */ block sits between the annotations and the method, so it isn't treated as Javadoc. Move it above @Test or make it a line comment.
4. Test coverage. The issue's reproducer is a record, which can't be written in core's Java 11 tests. A record case in testsuite/data (next to RecordSchemaTest) and a prefix-collision case in core would round out the coverage.