Ensure inlined method classes are dependencies

[cjjdespres]

No description provided.

[cjjdespres]

Attn @mpirvu. This does not seem to affect the failure rate much, if at all, because these classes will generally be mentioned in other relo records and so be captured as dependencies there. It's still good to make sure they're counted, I think.

[mpirvu]

LGTM

[mpirvu]

jenkins test sanity all jdk21

[mpirvu]

AIX compilation error:

16:06:08  /home/jenkins/workspace/Build_JDK21_ppc64_aix_Personal/openj9/runtime/compiler/p/codegen/J9AheadOfTimeCompile.cpp:254:118: error: too few arguments to function call, expected 5, have 4
16:06:08           acaRecord->setClassChainForInlinedMethod(reloTarget, classChainOffsetInSharedCache, self(), classChainRecord);
16:06:08           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~                                                                    ^
16:06:08  /home/jenkins/workspace/Build_JDK21_ppc64_aix_Personal/openj9/runtime/compiler/runtime/RelocationRecord.hpp:1811:7: note: 'setClassChainForInlinedMethod' declared here
16:06:08        void setClassChainForInlinedMethod(TR_RelocationTarget *reloTarget, uintptr_t classChainOffsetInSharedCache,
16:06:08        ^

[cjjdespres]

I'll fix that. I also see that I missed adding a dependency in TR_RelocationRecordPointer::setClassChainForInlinedMethod, which is distinct from TR_RelocationRecordProfiledInlinedMethod::setClassChainForInlinedMethod. That should be the last place.

I tested #20770 with this change added to make sure the failures weren't a result of the lack of dependencies. I'll double-check that PR with this last dependency added.

[cjjdespres]

This should be fixed. Adding that last dependency did not seem to affect anything.

[mpirvu]

jenkins test sanity all jdk21