Reorganize semantic tests for calldata and abicoder

[cameel]

This is just a refactor in preparation for fixing a bug in calldata validation. I needed to figure out what we have coverage for and since our tests for this are all over the place, I did some renaming and reorganization of the test dirs.

Summary of changes:

• abiEncodeDecode/, abiEncoderV1/, abiEncoderV2/, calldata/ under semanticTests/ all had random subsets of ABI encoding and calldata tests. I merged them all into a single abicoder/ dir with subdirs for validation, cleanup, structs, arrays, etc. Also moved some tests from arrays/ and structs/ there.
• This revealed that some of the tests are almost exact duplicates. I removed these.
  • Note that tests without the abicoder pragma cover both v1 and v2 since isoltest inserts a pragma depending on whether it runs with or without --abiencoderv1. Those with v2 pragma cover only v2. Those with v1 pragma cover v1 via evmasm and v2 via Yul (since the compiler ignores the pragma via Yul).
  • There are also quite a few non-identical tests covering the same thing. I intentionally left those, as I did not want to get into figuring out which version is better and should be left in.
• I enabled debug revert strings in tests covering validations. Otherwise it's not clear which validation failed in the test. Sometimes that required creating separate v1 and v2 versions.
• Other minor corrections: one test no longer needs compileViaYul: false. Another mixed tabs and spaces. Etc.

[cameel]

copy_from_calldata_removes_bytes_data.sol is the only test I left under calldata/. It does not really test decoding, but I'm also not sure what it really tests. It comes from #12289, which extracted tests from SolidityEndToEndTest.cpp, but this does not seem to be one of them. It's as if it appeared out of nowhere. It's even weirder that I reviewed that PR and apparently did see the original back then. I suspect that something got botched when the PR was rebased.

In any case, this test could be covering some ancient, long fixed bug, but in isolation it looks a bit worthless and weird (why unused set() function?) so I'm considering just removing it.

[rodiazet]

LGTM. Some comments or rather questions added and I proposed to move one test.

[rodiazet]

This test is a copy of test/libsolidity/semanticTests/array/calldata_array_bounds_check_nested_bytes.sol Shouldn't this be moved to test/libsolidity/semanticTests/abicoder/calldataDecoding/array

[cameel]

This test is a copy of test/libsolidity/semanticTests/array/calldata_array_bounds_check_nested_bytes.sol

Yes, that's why I kept only one of them.

Shouldn't this be moved to test/libsolidity/semanticTests/abicoder/calldataDecoding/array

Well, we have many tests that would fit in more than one place. When it comes to calldata arrays, for example, there is a big overlap between array behavior and calldata decoding. This case IMO is more about the former. The test checks that accessing an element past the end of the array reverts. This is not specific to calldata and happens with all kinds of arrays (note that it's Panic, not Error).

[cameel]

copy_from_calldata_removes_bytes_data.sol is the only test I left under calldata/.

Still not sure what to do about it, but for now I decided to move it to abicoder/calldataDecoding/ to avoid having an almost empty calldata/ dir. If we keep it, things will get messy again, because people will start adding new tests there.

[rodiazet]

LGTM