Corrected BLP1 alpha depth handling

[radarhere]

Resolves #8650

https://www.hiveworkshop.com/threads/blp-specifications-wc3.279306/ describes the start of the 'BLP File Header' as

uint32_t magic;
uint32_t content;
if (version >= 2) {
    uint8_t encodingType; // not documented
    uint8_t alphaBits;
    uint8_t sampleType; // not documented
    uint8_t hasMipmaps;
} else {
    uint32_t alphaBits;
}

However, we don't currently perform different behaviour for BLP1, and only treat the data as if it was BLP2.

Pillow/src/PIL/BlpImagePlugin.py

Lines 300 to 303 in aa0f412

	(self._blp_encoding,) = struct.unpack("<b", self._safe_read(1))
	(self._blp_alpha_depth,) = struct.unpack("<b", self._safe_read(1))
	(self._blp_alpha_encoding,) = struct.unpack("<b", self._safe_read(1))
	self.fd.seek(1, os.SEEK_CUR) # mips

This PR fixes updates how the header data is handled for BLP1, for both reading and writing.

[sminjard]

Wow, I didn't expect to hear from you this soon.
Thanks!

Everything seems fine. A bit confused about why you had to change things in the _openfunction. I think there is a way to call _read_blp_header so that you don't have that logic at two places, but this will go in another PR as this one seems to resolve the issue.

Happy New Year !

[radarhere]

Thanks for pointing out the duplicate code. As you can see, I've pushed a commit to fix it.

I can see you've approved this PR three times in the last ten minutes. I imagine you're experiencing a bug on your device and not seeing that it worked? If so, I can confirm you have approved this.

[sminjard]

Somehow GitHub was not checking the "#approved-reviews-by>=1" for auto merge check.

[hugovk]

This was a manual merge by me, so it didn't use automerge, which would have also needed the label adding and approval by a committer.