Improve method selector on contracts

[xunilrj]

Description

This PR continues the optimisation of contract calls on top of #7455. Now the contract method selector is optimised. Ideally, the method selection would just be a match and we would let the compiler generate the best code.

But we are not there yet. So, for now, we are going to first test the called method length. And based on its length we are going to check the whole string. Something like:

if called_method.len() == 5 {
    if called_method == "abcde" {
        ...
    }
    if called_method == "fghij" {
        ...
    }
}

if called_method.len() == 10 {
    if called_method == "..." {
        ....
    }
}

This alone reduces gas a lot.
To understand why match_expressions_all gas usage is much worse see below.

Test	Before	After	Percentage
should_pass/empty_fields_in_storage_struct (test.toml)::test_read_write_bytes	17533	12080	31.10%
should_pass/empty_fields_in_storage_struct (test.toml)::test_read_write_map	13986	8502	39.21%
should_pass/empty_fields_in_storage_struct (test.toml)::test_read_write_vec	29288	12797	56.31%
should_pass/language/associated_const_abi (test.toml)::test	7817	4547	41.83%
should_pass/language/associated_const_abi_multiple (test.toml)::test	2097	1511	27.94%
should_pass/language/associated_const_in_decls_of_other_constants (test.toml)::test	709	537	24.26%
should_pass/language/contract_ret_intrinsic (test.toml)::test	1809	1465	19.02%
should_pass/language/match_expressions_all (test.toml)	764	3708	-385.34%
should_pass/language/pusha_popa_multiple_defreg (test.toml)::incorrect_pusha_popa	624	451	27.72%
should_pass/language/raw_identifiers (test.error_type.toml)::test	1075	903	16.00%
should_pass/language/raw_identifiers (test.toml)::test	1075	903	16.00%
should_pass/language/references/mutability_of_references_memcpy_bug (test.toml)::test	1203	1031	14.30%
should_pass/language/slice/slice_contract (test.toml)::test_success	1145	920	19.65%
should_pass/language/string_slice/string_slice_contract (test.toml)::test_success	1322	1079	18.38%
should_pass/stdlib/storage_vec_insert (test.toml)::test_test_function	3831	3669	4.23%
should_pass/storage_element_key_modification (test.toml)::test_storage_key_address	1138	943	17.14%
should_pass/storage_element_key_modification (test.toml)::test_storage_key_modification	757	465	38.57%
should_pass/storage_slot_key_calculation (test.toml)::test	2979	2785	6.51%
should_pass/superabi_contract_calls (test.toml)::tests	1812	1310	27.70%
should_pass/superabi_supertrait_external_call (test.toml)	95	76	20.00%
should_pass/superabi_supertrait_same_methods (test.toml)::tests	936	634	32.26%
should_pass/test_abis/abi_impl_methods_callable (test.toml)::tests	772	599	22.41%
should_pass/test_abis/contract_abi-auto_impl (test.toml)::tests	772	599	22.41%
should_pass/unit_tests/aggr_indexing (test.toml)::test1	5443	4282	21.33%
should_pass/unit_tests/contract-multi-contract-calls (test.toml)::test_contract_2_call	804	631	21.52%
should_pass/unit_tests/contract-multi-contract-calls (test.toml)::test_contract_call	807	634	21.44%
should_pass/unit_tests/contract-multi-contract-calls (test.toml)::test_contract_multi_call	1592	1246	21.73%
should_pass/unit_tests/contract_multi_test (test.toml)::test_fail	808	635	21.41%
should_pass/unit_tests/contract_multi_test (test.toml)::test_success	806	633	21.46%
should_pass/unit_tests/script-contract-calls (test.toml)::test_contract_call	736	563	23.51%
should_pass/unit_tests/workspace_test (test.toml)::test_fail	808	635	21.41%
should_pass/unit_tests/workspace_test (test.toml)::test_success	807	634	21.44%

Radix Trie

This PR also turns off the radix trie optimisation at sway-core/src/semantic_analysis/ast_node/expression/match_expression/typed/typed_match_expression.rs. I am not 100% sure of its correctness and heuristics.

We need a better way to decide when not to use it. One of the examples where the generated code is worse is when all strings are very similar. For example: get_a , get_b. The generated trie will first test its length. Both have length 5. Not much was gained. Then it tests if for get_, which makes sense as it is the common substring. And the last step will test for a or b.

The issue is that all branches and jumps of the last step are not worthy to test just one character. Is probably cheaper to just test the whole string for each option.

In the end, we need a better heuristic to determine when to use this optimisation.

Checklist

• I have linked to any relevant issues.
• I have commented my code, particularly in hard-to-understand areas.
• I have updated the documentation where relevant (API docs, the reference, and the Sway book).
  • If my change requires substantial documentation changes, I have requested support from the DevRel team
• I have added tests that prove my fix is effective or that my feature works.
• I have added (or requested a maintainer to add) the necessary Breaking* or New Feature labels where relevant.
• I have done my best to ensure that my PR adheres to the Fuel Labs Code Review Standards.
• I have requested a review from the relevant team or maintainers.

[codspeed-hq]

CodSpeed Performance Report

Merging #7458 will not alter performance

_{Comparing xunilrj/improve-contract-call-3 (7a1f2ea) with master (e6f2ac8)}

Summary

✅ 25 untouched