feat(agent): ClassFile API backend ordinary args + AnyType[] + review-nit cleanup

[jbachorik]

What does this change do?

Follow-up to #843 (the ClassFile API backend parity work, squash-merged into develop as 9f39a250). Closes the parity gaps #843 left open and sweeps the open code-quality review findings:

• ENTRY/RETURN ordinary method arguments — handlers declaring ordinary enclosing-method args (typed or AnyType[]) were silently skipped by the ClassFile API backend. Added fixed-local-slot loading (methodArgSlot) threaded via new ENTRY/RETURN emitProbeCall overloads.
• AnyType[] aggregate packaging (ENTRY/RETURN/CALL) — added emitAnyTypeArray (anewarray + per-element load/box/arrayStore, mirroring ASM AnyTypeArgProvider), detected via the pre-replacement descriptor; plus an AnyType[] short-circuit in canEmitCallProbe (without it sameStackType rejected the Object[] aggregate and the probe was silently skipped).
• ClassFilter.isSubTypeOf — folded the JLS-dead if (cache == null) guard into the ci == null conservative no-match path (ClassCache.getInstance() uses the holder idiom, never null under normal class-init semantics).
• ClassFileApiFeatureSmokeTest — referenced each previously-unused captured probe param via BTraceUtils.str() / array.length, clearing the "useless parameter" findings and strengthening the smoke assertions.
• canEmitCallProbe — removed an unread staticCall local.
• Plan doc at docs/superpowers/plans/2026-07-06-classfileapi-remaining-gaps.md documenting the gaps, the ASM parity references, and per-task status.

Related issue

N/A — follow-up to #843.

Scope and compatibility

• I identified the affected module(s) and kept unrelated changes out of this PR.
• This preserves the supported Java/runtime compatibility tiers, or the change is documented below.
• This does not change the masked-JAR layout, class-loader boundary, or wire protocol.
• If it does, I updated the relevant architecture documentation and verification plan.

Compatibility or migration notes: None. ClassFile API backend changes only affect Java 26+ (class-file major ≥ 70) instrumentation; the ASM backend path is untouched. No API/CLI/protocol/packaging change.

Testing

• Unit tests
• Integration tests (if applicable)
• spotlessCheck
• Documentation/link or sample verification (if applicable)

Commands and results:

# Unit + verifier (JDK 26)
JAVA_HOME=$HOME/.sdkman/candidates/java/26-tem GRADLE_USER_HOME=$(pwd)/.gradle-user \
  ./gradlew :btrace-agent:classFileApiBackendTest
# -> 121 tests, 0 failures, 0 skipped

# Full agent suite
GRADLE_USER_HOME=$(pwd)/.gradle-user ./gradlew :btrace-agent:test
# -> BUILD SUCCESSFUL

# Formatting
JAVA_HOME=$HOME/.sdkman/candidates/java/26-tem GRADLE_USER_HOME=$(pwd)/.gradle-user \
  ./gradlew :btrace-agent:spotlessApply :integration-tests:spotlessApply
# -> BUILD SUCCESSFUL

# Integration matrix (-Pintegration test), dist btrace.jar v3.0.0-SNAPSHOT
# JDK 8/11/17(zulu)/21/26 -> BUILD SUCCESSFUL
# BTraceFunctionalTests.testOnMethod() PASSED on 8/11/17/21 (ClassCache-init race path)
# ClassFileApiTests smoke (Task 8 str() changes) PASSED on JDK 26
# JDK 17.0.19-sem: 6 PreparedModeAuthenticationFunctionalTest failures with
#   AttachNotSupportedException — JDK-build attach-infra issue, not this branch
#   (zulu-17.0.18 passes the same auth tests).

Documentation and release impact

• User-facing documentation is updated, or no documentation change is needed.
• Release notes/changelog are updated, or no release-note entry is needed.
• Samples, distribution contents, or published coordinates are updated if affected.
• This change is safe to merge independently of a release, or the dependency is explained below.

Release notes / follow-up work: No user-facing behavior change beyond fixing silently-skipped probes on Java 26+. Known out-of-scope edge case: sub-int (byte/short/char/boolean) AnyType[] elements box via the existing boxPrimitiveReturn (consistent with the @Return boxing and the JVMS verifier's int-collapse); not exercised by tests (ASM AnytypeArgs likewise covers only String/long/String[]/int[]). A deferred Phase-1 shared-validation refactor (the canEmit* duplication) is intentionally NOT included — pure refactor, no functional gain, integration just proved the current code works.

Final checklist

• I reviewed the complete diff and removed unrelated changes.
• New or changed behavior has appropriate tests, or the reason for not adding them is explained above.
• User-facing behavior, APIs, samples, and documentation are consistent with this change.
• I did not include generated build output, local configuration, credentials, or other accidental files.

---

This change is