`--max-size` or `--scale` ignores `--fix-size`, outputs images with width/height not multiple of 4

[Nathan8489]

Describe the bug
When using --max-size or --scale together with --fix-size, the output image width or height is not aligned to a multiple of 4 as expected.

To Reproduce
Steps to reproduce the behavior:

1. Add --fix-size or -fs to arguments.
2. Add --max-size 10 to arguments.
3. Use any PNG file as input.
4. Run the command. The generated DDS file will have a width or height not divisible by 4 (e.g. 10x10).

Expected behavior
The output image should have both width and height as multiples of 4.
This is required for BC compression formats and should be ensured by --fix-size.

System (please complete the following information):
Not relevant.

Additional context
fix_size is only used in filter_decode_png.cpp and not applied in filter_scale_image.cpp.

todds/src/pipeline/filter_decode_png.cpp

Lines 68 to 73 in ca6c521

	if (_fix_size && (first.width() % 4 != 0 || first.height() % 4 != 0)) [[unlikely]] {
	result = fix_image_size(*result, _mipmaps);
	const auto& new_first = result->get_image(0UL);
	file_data.width = new_first.width();
	file_data.height = new_first.height();
	}

todds/src/pipeline/filter_scale_image.cpp

Lines 37 to 56 in ca6c521

	auto width = (input_image.width() * _scale) / 100U;
	auto height = (input_image.height() * _scale) / 100U;
	if (_max_size > 0U && (width > _max_size || height > _max_size)) {
	if (width > height) {
	const double ratio = static_cast<double>(_max_size) / static_cast<double>(width);
	width = _max_size;
	height = static_cast<std::size_t>(static_cast<double>(height) * ratio);
	} else {
	const double ratio = static_cast<double>(_max_size) / static_cast<double>(height);
	height = _max_size;
	width = static_cast<std::size_t>(static_cast<double>(width) * ratio);
	}
	}
	if (width == 0 || height == 0) {
	_updates.emplace(report_type::pipeline_error,
	fmt::format("Could not scale {:s} from ({:d}, {:d}) to ({:d}, {:d}).", _paths[img->file_index()].first.string(),
	input_image.width(), input_image.height(), width, height));
	return nullptr;
	}