Reduce Waddy RAM usage

[khanghugo]

Right now, it should use 2*n RAM because

1. Waddy instance has WAD file data and
2. texture data for rendering

For some reasons, it is 4*n now. Maybe impl Drop?

[khanghugo]

It seems like this is where the issue comes from

gchimp/src/gui/programs/waddy.rs

Lines 879 to 900 in 4e6a372

	let texture_tiles = waddy
	.wad()
	.entries
	.iter()
	.enumerate()
	.filter_map(|(index, entry)| {
	if let FileEntry::MipTex(miptex) = &entry.file_entry {
	let wad_image = WadImage::from_wad_image(
	ui,
	entry.directory_entry.texture_name.get_string(),
	miptex.mip_images[0].data.get_bytes(),
	miptex.palette.get_bytes(),
	(miptex.width, miptex.height),
	);
	return Some(TextureTile::new(index, wad_image));
	// None
	}
	None
	})
	.collect::<Vec<TextureTile>>();

egui::TextureHandle does have impl Drop and GPU memory does decrease when the image is dropped. But it still uses too much.

[khanghugo]

After some investigations, I have missed something fairly simple, the textures are stored as RGBA8 in memory. For a 1000 512x512 WAD file, it requires at least 750MB of RAM to store them. So I guess the behavior should be expected.

Recently, I have tried out iced and have a WIP Waddy. It seems like it uses a lot less RAM. One thing that I have overlooked is VRAM usage and iced version does use less VRAM as well.

A possible rewrite of gchimp-native under iced is considered.