Logical with Immediate meta-mnemonics and and bug fixes

[burhanr13]

I have added meta mnemonics for logical operations with immediate (and_imm, ands_imm,orr_imm,eor_imm) and a function to check whether a value is a valid immediate for these kinds of instructions. I also fixed align() to not add padding if curr is already aligned, and reset() to reset the protection mode to RW so the code can be written to again.

[kawakami-k]

Hi, @burhanr13 !
Thank you very much for you contribution! I'll check the PR and merge soon!

[kawakami-k]

@herumi

I think this PR is fine to merge. Would you please check this PR too? If there are no problems, please merge it. After that, I will add test patterns for this PR for CI and update the version number.

[herumi]

It's good for me, but if add_imm is changed to

void and_(const WReg &dst, const WReg &src, uint32_t imm, const WReg *tmp = 0) {
  if (isValidLogicalImm(imm, 32)) {
    // original and_(dst, src, imm);
  } else {
    mov(tmp, imm);
    // original and_(dst, src, tmp);
  }
}

then we can use the same function name. How do you think about it?

[kawakami-k]

@herumi

Although it is not a clear policy, the current xbyak_aarch64_mnemonic.h contains the implementations of the instructions that exist in Armv8-v9 instruction set, and the pseudo-instructions unique to xbyak_aarch64 are in xbyak_aarch64_meta_mnemonic.h. To maintain this consistency, it would be good to name the pseudo-instructions like and_imm and put them in `xbyak_aarch64_meta_mnemonic.h".

[herumi]

@kawakami-k
I see, it's okay.

[kawakami-k]

@burhanr13
Thank you for the contribution!