Metadata for every format

[matmen]

All encoders now take the metadata encode() takes for PNGs. The interface is now called Metadata, PNGMetadata stays as a deprecated alias until 4.0.

Each format stores the fields it has a place for:

Format	Stored in	Fields
PNG, APNG	text chunks	all
JPEG	Exif (APP1), comment marker	author, description, copyright, creation time, software, comment
WebP (still and animated)	Exif (EXIF chunk, extended format)	author, description, copyright, creation time, software
GIF	comment extensions	software, comment
TGA	TGA 2.0 extension area	author, comment, creation time, software

Defaults

The defaults are the same for every format: the creation time, and ImageScript with its version as the software. {software: ''} leaves the software out.

Size added to each file: about 40 bytes for GIF, about 200 for JPEG and WebP, 495 for TGA (the extension area has a fixed size).

Fixes

• PNG text that isn't Latin-1: it was written as UTF-8 into tEXt, which is Latin-1, so readers showed it garbled. It now goes into iTXt chunks as UTF-8.
• PNG text chunks now come before the image data, where all readers look for them.
• An invalid creationTime throws a RangeError instead of writing "Invalid Date".

Testing

• New test: tests/metadata.js.
• ExifTool 13.59 reads the JPEG, WebP, GIF, PNG and APNG fields without warnings.
• Pillow reads all fields. Non-ASCII Exif text shows garbled there, because Pillow decodes Exif strings as Latin-1. ExifTool shows it correctly, and UTF-8 is the common practice for Exif strings.
• TGA: ExifTool doesn't support TGA. The extension area was parsed by hand against the TGA 2.0 specification.