Missing kerning for single-byte strings in PDF #19581

Bug report

Bug summary
Kerning is not applied in the PDF output of pure single-byte strings which leads to inconsistencies between PDF and other output formats.

Code for reproduction

import matplotlib.pyplot as plt

fig, ax = plt.subplots()
ax.text(0.1, 0.5, "ATL", size=120)
fig.savefig("atl.pdf")
fig.savefig("atl.png", dpi=300)

Actual outcome

Expected outcome
I expected that the spacing between A and T is consistent between both formats. Besides this minor visual issues, I use get_window_extent() to place a second text item in bold next to the first one. The space between the two texts varies significantly between output formats. get_window_extent() considers the kerning information, so the space looks correct in PNG but is usually too small in PDF files.

The issue is related to #18282. Here the same issue was fixed for strings with multi-byte chars. I don't see the changes from #18282 in version 3.3.4, but they will not effect the single-byte string case.

Unaware of #18282, I've implemented a fix for the single-byte case. I will open a PR, but you might have other ideas on how to implementation a fix.

Matplotlib version

• Operating system: Ubuntu 16.04
• Matplotlib version: 3.3.4
• Matplotlib backend: TkAgg (but actually PDF?)
• Python version: 3.7.5
• Font: DejaVuSans Type 3

I've installed matplotlib in a virtualenv with pip.