RFR: 8258015: [JVMCI] JVMCI_lock shouldn't be held while initializing box classes for AOT

[Doug Simon]

This PR fixes a regression caused by JDK-8257917 by not locking JVMCI_lock when initializing the boxing cache classes for AOT.

It also reduces the calls to `JVMCI::ensure_box_caches_initialized` to be at most one per JVMCI code installation.

-------------

Commit messages:
 - only call JVMCI::ensure_box_caches_initialized at most once per JVMCI code installation
 - do not lock JVMCI_lock when initializing boxing cache classes for AOT

Changes: https://git.openjdk.java.net/jdk/pull/1730/files
 Webrev: https://webrevs.openjdk.java.net/?repo=jdk&pr=1730&range=00
  Issue: https://bugs.openjdk.java.net/browse/JDK-8258015
  Stats: 27 lines in 5 files changed: 10 ins; 8 del; 9 mod
  Patch: https://git.openjdk.java.net/jdk/pull/1730.diff
  Fetch: git fetch https://git.openjdk.java.net/jdk pull/1730/head:pull/1730

PR: https://git.openjdk.java.net/jdk/pull/1730