feat!: add HitResultPriority::Fastest variant (#58)

* feat!: add HitResultPriority::Fastest variant

* fix: osu std hitresult calc

* fix: consider slider value

* test: remove linux test cases

The underlying math lib on linux is inherently different for C#
and Rust so there is no point in asserting equality. Specifically,
Math.Cbrt(7512727.819502543) resulted in a different value than
the analogous f64::cbrt, i.e. 195.854047766587 in C# vs
195.85404776658697 in Rust).

Author: MaxOhn

src/any/performance/mod.rs

@@ -434,11 +434,14 @@ impl<'map> Performance<'map> {
 
 /// While generating remaining hitresults, decide how they should be distributed.
 #[derive(Copy, Clone, Debug, Eq, PartialEq)]
+#[non_exhaustive]
 pub enum HitResultPriority {
     /// Prioritize good hitresults over bad ones
     BestCase,
     /// Prioritize bad hitresults over good ones
     WorstCase,
+    /// Prioritize fast hitresults generation
+    Fastest,
 }
 
 impl HitResultPriority {

src/catch/performance/mod.rs

@@ -347,6 +347,8 @@ impl<'map> CatchPerformance<'map> {
             let max_tiny_droplets =
                 cmp::min(attrs.n_tiny_droplets, raw_tiny_droplets.ceil() as u32);
 
+            // Hopefully using `HitResultPriority::Fastest` wouldn't make a big
+            // difference here so let's be lazy and ignore it
             for n_tiny_droplets in min_tiny_droplets..=max_tiny_droplets {
                 let n_tiny_droplet_misses = attrs.n_tiny_droplets - n_tiny_droplets;