Re-work stable_partition to avoid self-comparison (#33)

* Update sort-test-tools to include self_cmp tests

* Re-work stable_partition to avoid self-comparison

- binary-size -> improvement, especially for opt=s and string
- compile-time -> minor regression for release 8s -> 8.3s
- run-time -> minor regression 2-5% for i32,u64,1k 5-8% improvement for
  string and 18% regression for f128

* Apply review comments

Cargo.toml

@@ -15,7 +15,7 @@ test = false
 [dependencies]
 
 [dev-dependencies]
-sort_test_tools = { git = "https://github.com/Voultapher/sort-research-rs", rev = "ed65ed800666c1fb06d93d5f31c8d0add55bf32e", default-features = false }
+sort_test_tools = { git = "https://github.com/Voultapher/sort-research-rs", rev = "9cd351e9e6ded91a45d20c63d66b47f104cad527", default-features = false }
 
 [features]
 default = ["large_test_sizes"]