[ybai08]

What kind of change does this PR introduce?

Bug fix. Fixes #1425, also fixes #606 (and probably #1354, which uses the default Helvetica and a template literal with Windows line endings).

With a standard (AFM) font, AFMFont#encodeText writes each character code with char.toString(16), so a code below 0x10 comes out as a single hex digit. In the <...> string the following digits then pair up wrongly. Per PDF 32000-1 §7.3.4.3 a reader takes the digits two at a time, and assumes a 0 after a trailing odd digit.

• doc.text('line one\r\nline two'): the first line is written as <6c696e65206f6e65d>. The final d is read as 0xD0, which is Ð in WinAnsiEncoding. That is the stray Ð reported in Extra "Ð" at the end of a line when using text with carriage returns (\r) #1425 and Strange character "Ð" every time I press "intro" #606.
• A tab inside a line damages everything after it. doc.text('a\tb', { lineBreak: false }) is written as <61962>, i.e. bytes 61 96 20, so the b is lost. doc.text('Tab\there', { lineBreak: false }) is read back by pdf.js as Tab–†W&P.

The fix pads each code to two digits (padStart(2, '0')), as the embedded font path already does with its 4-digit glyph IDs. Codes 0x10 and above are unchanged, so normal text produces the same bytes (the visual snapshots still pass).

Text extracted with pdfjs-dist (already a dev dependency), doc.text(s, 50, 50, { lineBreak: false }):

input	before	after
'a\tb'	a–	a b
'Tab\there'	Tab–†W&P	Tab here
'line\r'	lineÐ	line
'x\u0001yz'	x —	x yz

Not part of this PR: the line wrapper still keeps the \r of a \r\n at the end of the line (now as an invisible 0x0D code instead of Ð). Stripping it the way \n is stripped would be a separate change.

Checklist:

• Unit Tests
• Documentation N/A
• Update CHANGELOG.md
• Ready to be merged

Tests added:

• tests/unit/afm.spec.js: encodeText('a\tb\r\x01') returns ['61', '09', '62', '0d', '01']
• tests/unit/text.spec.js: doc.text('a\tb\rc', { lineBreak: false }) keeps the text a\tb\rc in the content stream

Both fail on master and pass with the fix. Ran yarn test (62 files, 552 tests, unit and visual), yarn lint and yarn prettier, all passing. Did not run test:tools or test:package.