aptos-core [compiler-v2] Fix evaluation order to be consistent in language version >= 2

Description

Major changes in this PR:

we showcase that the evaluation order used by compiler v1 in the presence of sequences intermixed with binary operators is hard to explain and understand
thus, we make the decision in compiler v2 to:
- report a compiler error if language version is below 2.0 and there could be divergence in semantics of the program (trying to simulate the exact v1 behavior is error-prone due to a variety of reasons); this should have no false negatives, but could have false positives which in some cases can be addressed with additional complexity
- diverge the semantics (breaking change) if the language version is >= 2.0, by having consistent left-to-right evaluation order throughout, for operators and functions alike
fixes a problem with non-binary operators and evaluation order, such as for vector pack instructions (eg, this example is now fixed: https://github.com/aptos-labs/aptos-core/issues/11447#issuecomment-2008278801)

Minor changes in this PR:

transactional tests now have include/exclude options in test configs
binary operator to string conversion is canonicalized and code elsewhere is updated to use this, causing minor changes in test outputs
code duplication reducing refactors

Evaluation order in compiler v1

Consider the following examples to see some samples of evaluation order used by compiler v1 in the presence of sequences within the context of binary operations. They are meant to showcase how concisely describing the v1 ordering is hard (as opposed to, a left-to-right evaluation ordering everywhere).

These are also present at the module documentation for third_party/move/move-compiler-v2/src/seqs_in_binop_checker.rs. I am re-presenting them for an overview to bolster my argument about decisions made in this PR.

We number the sub-expressions in their order of their evaluation. Some (sub-)expressions are left un-numbered if they are irrelevant to the understanding of the evaluation order.

case 1: add is a user-defined function.

let x = 1;
add({x = x - 1; x + 8}, {x = x + 3; x - 3}) + {x = x * 2; x * 2}
     ^^^^^^^^^  ^^^^^    ^^^^^^^^^  ^^^^^      ^^^^^^^^^  ^^^^^
        |        |         |          |            |        |
        |        |         |          |            |        |
        1        |         |          |            |        |
                 2         |          |            |        |
                           3          |            |        |
                                      |            4        |
                                      5                     |
                                                            6

case 2:

fun aborter(x: u64): u64 {
    abort x
}

public fun test(): u64 {
    let x = 1;
    aborter(x) + {x = x + 1; aborter(x + 100); x} + x
    ^^^^^^^^^^    ^^^^^^^^^  ^^^^^^^^^^^^^^^^
       |              |              |
       |              1              |
       |                             2
    never evaluated
}

case 3:

(abort 0) + {(abort 14); 0} + 0
 ^^^^^^^      ^^^^^^^^
    |              |
    1              |
                never evaluated

case 4:

{250u8 + 50u8} + {abort 55; 5u8}
 ^^^^^^^^^^^^     ^^^^^^^^
     |               |
     |               1
  never evaluated

case 5:

let x = 1;
x + {x = x + 1; x} + {x = x + 1; x}
^    ^^^^^^^^^  ^     ^^^^^^^^^  ^
|       |       |        |       |
|       1       |        |       |
|               |        2       |
3               3                3

Type of Change

[x] Bug fix
[x] Breaking change

Which Components or Systems Does This Change Impact?

[x] Move/Aptos Virtual Machine

How Has This Been Tested?

added around 52 new tests that expose subtle semantic behavior and cover previously untested parts of the language
- in each case, I have manually verified that whenever there is a divergence in semantics (test outputs differ) between v1 and v2 (with lang v2), we report an error when using "compiler v2 + lang v1"
previous tests pass, one affected test has been moved to a new folder, many baselines have been updated but verified

Apr 18 '24 17:04 vineethk

⏱️ 6h 33m total CI duration on this PR

Job	Cumulative Duration	Recent Runs
rust-move-tests	1h 44m	🟥 🟥 🟥 🟥 🟥 (+4 more)
rust-targeted-unit-tests	1h 39m	🟥 ⬜ 🟥 🟩 🟩 (+1 more)
rust-move-unit-coverage	1h 36m	🟩 ⬜ 🟩 🟩 🟩 (+1 more)
rust-lints	43m	🟥 🟥 🟩 🟩 🟩 (+1 more)
run-tests-main-branch	25m	🟩 🟩 🟩 🟩 🟩 (+1 more)
check-dynamic-deps	11m	🟩 🟩 🟩 🟩 🟩 (+1 more)
general-lints	9m	🟩 🟩 🟩 🟩 🟩 (+1 more)
semgrep/ci	2m	🟩 🟩 🟩 🟩 🟩 (+1 more)
file_change_determinator	1m	🟩 🟩 🟩 🟩 🟩 (+1 more)
file_change_determinator	1m	🟩 🟩 🟩 🟩 🟩 (+1 more)
permission-check	18s	🟩 🟩 🟩 🟩 🟩
permission-check	13s	🟩 🟩 🟩 🟩 🟩
permission-check	13s	🟩 🟩 🟩 🟩 🟩
permission-check	13s	🟩 🟩 🟩 🟩 🟩

🚨 1 job on the last run was significantly faster/slower than expected

Job	Duration	vs 7d avg	Delta
rust-lints	5m	7m

_{settings ⋅ feedback ⋅ docs ⋅ learn more about trunk.io}

Apr 18 '24 17:04 trunk-io[bot]

~~Marking this as draft to handle some false positives in "Aptos Move Test" tests in CI.~~

Apr 18 '24 20:04 vineethk

This PR is now ready for review.

Apr 23 '24 16:04 vineethk

Codecov Report

Attention: Patch coverage is 94.04762% with 10 lines in your changes are missing coverage. Please review.

Project coverage is 57.6%. Comparing base (1af48e9) to head (12ed7ed). Report is 3 commits behind head on main.

Files	Patch %	Lines
third_party/move/move-model/src/ast.rs	75.6%	9 Missing :warning:
...piler-v2/src/env_pipeline/seqs_in_binop_checker.rs	99.0%	1 Missing :warning:

Additional details and impacted files

@@           Coverage Diff            @@
##             main   #12936    +/-   ##
========================================
  Coverage    57.5%    57.6%            
========================================
  Files         833      834     +1     
  Lines      198121   198257   +136     
========================================
+ Hits       113999   114212   +213     
+ Misses      84122    84045    -77

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

Apr 24 '24 15:04 codecov[bot]

@brmataptos @wrwg PTAL, I believe I have addressed your comments.

Apr 29 '24 16:04 vineethk