I recently used rst-backticks to find and fix some PEPs (python/peps#1554) and am planning to add it as a pre-commit linter on the CI.

I also fixed some related RST problems (python/peps#1560) where inline code touches normal text. This sort of stuff is allowed in Markdown but not reStructuredText:

-``PyMem_Realloc()`` indirectly call``PyObject_Malloc()`` and
+``PyMem_Realloc()`` indirectly call ``PyObject_Malloc()`` and

-This PEP proposes that ``bytes`` and ``bytearray``gain an optimised
+This PEP proposes that ``bytes`` and ``bytearray`` gain an optimised

-Reading this we first see the``break``, which obviously applies to
+Reading this we first see the ``break``, which obviously applies to

-for using``long_description`` and a corresponding
+for using ``long_description`` and a corresponding

Guido suggested a linter for this too (python/peps#1560 (review)), and as I used grep to find these:

$ git grep "[A-Za-z0-9]\`\`[A-Za-z0-9]"
pep-0445.txt:``PyMem_Realloc()`` indirectly call``PyObject_Malloc()`` and
pep-0467.txt:This PEP proposes that ``bytes`` and ``bytearray``gain an optimised
pep-0548.rst:Reading this we first see the``break``, which obviously applies to
pep-0621.rst:for using``long_description`` and a corresponding

then pygrep-hooks makes a good fit!

I didn't find anything in the spec where "alphanumeric double-backticks alphanumeric" is allowed so I think it's safe to add:

• https://docutils.sourceforge.io/docs/ref/rst/restructuredtext.html#inline-markup-recognition-rules

Naming things: always hard, feedback welcome for improvements on rst-inline-touching-normal.

Tests: I added the four real problems I found in the PEPs, plus some simplified versions. Let me know if you'd prefer only the simplified versions.