Implement boundary check with hardware trap for interpreter

[wenyongh]

Motivation

Currently the memory access boundary check with hardware trap feature for 64-bit platform is only supported in AOT/JIT mode, we hope to enable the feature for interpreter mode (both classic-interp and fast-interp):

1. To improve the performance of interpreter Normally interpreter performance in 64-bit can be improved by 1% to 5% depending on cases.
2. To prepare for multi-tier compilation feature To tier-up from Fast JIT to LLVM JIT, as hw bound check is supported in LLVM JIT but unsupported in Fast JIT, we had better enable the feature for interpreter and Fast JIT, so as to tier-up smoothly.

Platforms to support

1. Linux 64-bit
2. MacOS 64-bit
3. Windows 64-bit