Calldata validation tests

[wechman]

closes #12922

[ekpyron]

Any reason for not adding // revertStrings: debug to all of them? If that doesn't cause issues, it may make it easy to properly check that this now catches all revert conditions.

[wechman]

Any reason for not adding // revertStrings: debug to all of them? If that doesn't cause issues, it may make it easy to properly check that this now catches all revert conditions.

@ekpyron The reason is an error message that keeps changing in consecutive test runs. An example:

// f(uint256[][]): 0x20, 2, 0x40, 0xa0, 2, 5, 6, 3, 7, 8 -> FAILURE, 3963877391197344453575983046348115674221700746820753546331534351508065746944, 862718293348820473429344482784628181556388621521298319395315527974912, 0x2b414249206465636f64696e673a20696e76616c69642063616c6c6461, 52639898083992983106342913290719799829523823861698573317707643453664495927296, 0x4100000000000000c820b8edba5500000100000001000000

I guess we may want to fix that, but I am not sure if we should do that in scope of this ticket.

[ekpyron]

Ah right, I vaguely remember that there's some legacy vs. via-IR differences with some of these cases - we should indeed try to fix that and align those eventually, but true, no need to burden this PR with that.

[ekpyron]

indentation, also below

[wechman]

A fixup commit delivered. Also, I delivered another one that removes compileViaYul: also option, because meanwhile it became a default.