[misc] fix blog post generator quirks (#18601)

While preparing the 1.15 release blog posts, I noticed a few quirks with
our maarkdown converter:

- We treat any `\n` followed by a capital letter as a new paragraph,
which can cause too many `<p>` tags to be inserted at times (a common
cause in this blog post was a line break followed by the word "PR").
- Using multiple consecutive backticks for an inline code section is
valid markdown (it's commonly used for strings where you need to include
single backticks, eg ``` ``a string with a single ` :)`` ```), but our
script was confused by this and generated lots of erroneous `<tt>`
sections where they didn't belong.
- Including a `#\d` in the middle of a word caused the script to assume
it was a PR that it should link. In this specific case, the changelog
contains several occurrences of `mypy_mypyc-wheels#<PR number>`, which
the script was stomping on.

This PR contains some minor tweaks for the blog post generation script
that attempt to address these quirks.

misc/gen_blog_post_html.py

@@ -95,7 +95,7 @@ def convert(src: str) -> str:
     h = re.sub(r"`\*\*`", "<tt>**</tt>", h)
 
     # Paragraphs
-    h = re.sub(r"\n([A-Z])", r"\n<p>\1", h)
+    h = re.sub(r"\n\n([A-Z])", r"\n\n<p>\1", h)
 
     # Bullet lists
     h = format_lists(h)
@@ -104,6 +104,7 @@ def convert(src: str) -> str:
     h = format_code(h)
 
     # Code fragments
+    h = re.sub(r"``([^`]+)``", r"<tt>\1</tt>", h)
     h = re.sub(r"`([^`]+)`", r"<tt>\1</tt>", h)
 
     # Remove **** noise
@@ -125,7 +126,9 @@ def convert(src: str) -> str:
         r'fixes issue <a href="https://github.com/python/mypy/issues/\1">\1</a>',
         h,
     )
-    h = re.sub(r"#([0-9]+)", r'PR <a href="https://github.com/python/mypy/pull/\1">\1</a>', h)
+    # Note the leading space to avoid stomping on strings that contain #\d in the middle (such as
+    # links to PRs in other repos)
+    h = re.sub(r" #([0-9]+)", r' PR <a href="https://github.com/python/mypy/pull/\1">\1</a>', h)
     h = re.sub(r"\) \(PR", ", PR", h)
 
     # Markdown links