Check string comparison size

[mmisono]

In the debug build, the following assertion failure occurs:

% sudo ./src/bpftrace -e 'BEGIN /comm == "aaaaaaaaaaaaaaaaaaaaaaaaa"/ {}'
bpftrace: /home/ubuntu/work/bpftrace/bpftrace/src/ast/irbuilderbpf.cpp:683: llvm::Value *bpftrace::ast::IRBuilderBPF::CreateStrncmp(llvm::Value *, llvm::Value *, bpftrace::AddrSpace, std::string, uint64_t, const bpftrace::location &, bool): Assertion `valp->getElementType()->isArrayTy() && valp->getElementType()->getArrayNumElements() >= n && valp->getElementType()->getArrayElementType() == getInt8Ty()' failed.

The problem part is https://github.com/iovisor/bpftrace/blob/a03b5c4bbb12a25345a2fdec78374963ff936131/src/ast/irbuilderbpf.cpp#L682
and the reason is that the the size of "comm" is 16, but the size of
string literal is bigger than that.

To fix the problem, check the string size in the semantic analyzer and
error if the size of string literal exeeds the that of the other. Now
the above exmaple becomes

% sudo ./src/bpftrace -e 'BEGIN /comm == "aaaaaaaaaaaaaaaaaaaaaaaaa"/ {}'
stdin:1:7-43: ERROR: The right string literal has the longer length of the size of the left (16 < 25)
BEGIN /comm == "aaaaaaaaaaaaaaaaaaaaaaaaa"/ {}
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Checklist

• Language changes are updated in docs/reference_guide.md
• User-visible and non-trivial changes updated in CHANGELOG.md
• The new behaviour is covered by tests

---

I wonder if this could cause a regression. I don't think so, but...

[mmisono]

Updated:

• Changed the semantic analyzer so that it emits a warning instead of making an error. Also, make the check simple.
• In the CreateStrcmp, skip comparison if the condition is always false. Also, fix assertion so that it does not assert that case.

[fbs]

Looks good.

Having a warning test might be useful tho, just to catch this case in a test somewhere.

[mmisono]

Added test for string comparisons