Update image to 0.24

[avitex]

An allocation has been removed for writing a PNG data-url in piet-svg.

Rather than going through DynamicImage::write_to, we just take the bytes and write directly with PngEncoder. This is due to write_to now requiring an io::Seek constraint which EncoderStringWriter does not implement.

Note PngEncoder::encode is now deprecated, with a note pointing to ImageEncoder::write_image instead. Note also that ImageEncoder::write_image expects image buffers with color types with 16-bit per channel or larger to be native endian, which differs to the semantics of PngEncoder::encode.

[cmyr]

Thanks! I haven't tested this out but it looks reasonable, so I'm happy to merge and clean up any potential issues as they arise. :)

[cmyr]

I'm not sure how feasible this is, but in theory if we could estimate the image's final b64 length then we could preallocate the buffer, as in:

Suggested change

	let mut writer = base64::write::EncoderStringWriter::from(
	String::from("data:image/png;base64,"),
	base64::STANDARD,
	);
	static PRELUDE: &str = "data:image/png;base64,"'
	let encoded_len = guess_encoded_len_somehow() + PRELUDE.len();
	let mut buf = String::with_capacity(encoded_len);
	buf.push_str(PRELUDE);
	let mut writer = base64::write::EncoderStringWriter::from(
	&mut buf,
	base64::STANDARD,
	);

This is totally unnecessary but I thought I'd point it out for fun. :)

[avitex]

That would be fun actually. I'm starting to have a look at druid for a slow burn project with a lot of yak shaving. Part of that will be starting to dive a little into the stack and helping out where I can.

I gave this a test with picture_2 and the pictures render the same to current state. I will note the PngEncoder (with either write_image/encode) has a slightly larger output. I didn't look too deep into it, was more looking for correct output.

[avitex]

I'll make such changes in a separate PR

[cmyr]

@avitex you may need to rebase this on master to pass CI, I've fixed an issue that was causing it to fail.

[avitex]

@cmyr done :)

[avitex]

@cmyr Pushed up a change to fix the previous run