[vectorIntrinsics] RFR: 8264579: Use NEON vector load/store to access small size memory [v2]

[Ningsheng Jian]

> For vector load/store with size 2/4/8/16, we can directly use NEON load/store instead of partial load/store with predicate. Since
> currently we are still using different registers (vReg/vecX/vecD) for SVE and NEON codegen, we have to write different rules for NEON and SVE, though they are the same physical regs (sharing lower parts).

Ningsheng Jian has updated the pull request incrementally with one additional commit since the last revision:

  Fix xor reduction predicate

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

Changes:
  - all: https://git.openjdk.java.net/panama-vector/pull/64/files
  - new: https://git.openjdk.java.net/panama-vector/pull/64/files/53d184b6..6bdba39b

Webrevs:
 - full: https://webrevs.openjdk.java.net/?repo=panama-vector&pr=64&range=01
 - incr: https://webrevs.openjdk.java.net/?repo=panama-vector&pr=64&range=00-01

  Stats: 18 lines in 2 files changed: 6 ins; 0 del; 12 mod
  Patch: https://git.openjdk.java.net/panama-vector/pull/64.diff
  Fetch: git fetch https://git.openjdk.java.net/panama-vector pull/64/head:pull/64

PR: https://git.openjdk.java.net/panama-vector/pull/64