[code-reflection] RFR: Experiments focused on injecting buffer tracking on statement boundaries

[Gary Frost]

At present we inject buffer tracking before and after iface mutations in the compute method. 

This works fine, because the resulting model converted to bytecode is valid. 

Sadly if we reify back to java code (for jextracted backend) we have invalid java code. 

Or if we use the model to create C99 compute code, we have to align the instrumentation on statement boundaries. 

This PR adds some low level helper code to find statement boundaries and help transforms inject on boundaries.

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

Commit messages:
 - More experimentation around injecting buffer tracking on statement boundaries

Changes: https://git.openjdk.org/babylon/pull/851/files
  Webrev: https://webrevs.openjdk.org/?repo=babylon&pr=851&range=00
  Stats: 431 lines in 16 files changed: 282 ins; 78 del; 71 mod
  Patch: https://git.openjdk.org/babylon/pull/851.diff
  Fetch: git fetch https://git.openjdk.org/babylon.git pull/851/head:pull/851

PR: https://git.openjdk.org/babylon/pull/851