wazevo(ssa): avoids allocation during natural order sort

[mathetake]

This starts using slices.SortFunc (thanks to @achille-roussel) to avoid
allocations during ssa.passSortSuccessors. As a result for zig stdlib compilation,
huge amount of allocations has been eliminated.

                                        │    old.txt    │             new.txt              │
                                        │     B/op      │     B/op       vs base           │
Zig/optimizing/Compile/test-opt.wasm-10   513.2Mi ± ∞ ¹   506.6Mi ± ∞ ¹  ~ (p=1.000 n=1) ²
¹ need >= 6 samples for confidence interval at level 0.95
² need >= 4 samples to detect a difference at alpha level 0.05

                                        │    old.txt    │             new.txt             │
                                        │   allocs/op   │  allocs/op    vs base           │
Zig/optimizing/Compile/test-opt.wasm-10   1201.3k ± ∞ ¹   781.8k ± ∞ ¹  ~ (p=1.000 n=1) ²
¹ need >= 6 samples for confidence interval at level 0.95
² need >= 4 samples to detect a difference at alpha level 0.05

[achille-roussel]

Should the function return zero when i.ReturnBlock() && j.ReturnBlock() ?

I honestly don't know if it matters, but it could confuse the sorting algorithm if f(i, j) => 1 but also f(j, i) => 1.

The comparison function has three states bascially for less than, equal, and greater than, which is slightly different from the "less" function that was historically used in the sort package.

[mathetake]

yeah it should return zero but it doesn't make any difference in the result at the end of the day. Would you mind raising a fix if you like? 🏃

[achille-roussel]

These things make me nervous, I've seen quick-sort turn into O(N^2) because the comparison function wasn't producing a coherent output.

Sending a PR 👍