Use absolute instead of canonicalize for better Windows path handling

[shssoichiro]

Closes #860

[Uranite]

It seems that the cachepath is still canonicalized

as a result, when using bestsource, the path for the cache is wrong (see how the path repeats, and cache.json is a folder)
C:\Users\redacted\Videos\osu!\osu! 2024.05.25 - 12.33.34.01\split\cache.json\Users\redacted\Videos\osu!\osu! 2024.05.25 - 12.33.34.01.mp4.0.bsindex

Edit: Seems like I was incorrect, even if the path is an absolute path, the behaviour will still be the same because according to bestsource's README it seems to be an intended behaviour

cachepath: The path where cache files are written. Note that the actual index files are written into subdirectories using based on the source location. Defaults to %LOCALAPPDATA% on Windows and ~/bsindex elsewhere.

[Uranite]

Anyway, here are my proposed change for this PR
By using to_absolute_path for cache_file we can safely remove use path_abs::PathAbs;
Also, bestsource has stopped generating json indexes since a long time now, so I changed it to bsindex (path still all weird)

diff --git a/av1an-core/src/vapoursynth.rs b/av1an-core/src/vapoursynth.rs
index 7c8484e..4a5f1f5 100644
--- a/av1an-core/src/vapoursynth.rs
+++ b/av1an-core/src/vapoursynth.rs
@@ -5,7 +5,6 @@ use std::path::{Path, PathBuf};
 
 use anyhow::{anyhow, bail};
 use once_cell::sync::Lazy;
-use path_abs::PathAbs;
 use std::process::Command;
 use vapoursynth::prelude::*;
 use vapoursynth::video_info::VideoInfo;
@@ -201,13 +200,13 @@ pub fn create_vs_file(
 
   let mut load_script = File::create(&load_script_path)?;
 
-  let cache_file = PathAbs::new(temp.join("split").join(format!(
+  let cache_file = to_absolute_path(&temp.join("split").join(format!(
     "cache.{}",
     match chunk_method {
       ChunkMethod::FFMS2 => "ffindex",
       ChunkMethod::LSMASH => "lwi",
       ChunkMethod::DGDECNV => "dgi",
-      ChunkMethod::BESTSOURCE => "json",
+      ChunkMethod::BESTSOURCE => "bsindex",
       _ => return Err(anyhow!("invalid chunk method")),
     }
   )))?;

[Uranite]

I can confirm that this PR fixes bestsource performance, my proposed changes is not necessary, but it's up to you if you want

[shssoichiro]

Seems like a good idea to include anyway. Unfortunately what is blocking this PR is that something seems broken with the select chunk method, unrelated to this change, and I haven't been able to track down what is causing it yet. The fact that it broke without any code changes, makes me think it may be related to an ffmpeg or other library update...