fix: `<url>` processing of markdown

[acmepjz]

This PR fixes <url> processing problem of markdown introduced by #157. It turns out that the problem occurred at Lean.Xml.instToStringElement which does not do escape. The code is copied from Lean.Xml.instToStringElement with escape code added. The added functions should be removed once https://github.com/leanprover/lean4/issues/4411 is fixed.

discussion: https://leanprover.zulipchat.com/#narrow/stream/113488-general/topic/Contribution.20to.20doc-gen4.3F/near/443537944

[acmepjz]

Example of the changes of Init/Conv.html:

• before:

  line 27: in "head position" until a constructor is exposed. ...
  line 31: ... Performs one step of "congruence", which takes a term and produces
  line 92: ... closes one conv goal "trivially", by using reflexivity
  line 146: by focusing on particular subexpressions.</p><p>See <https://lean-lang.org/theorem_proving_in_lean4/conv.html> for more details. ...

• after:

  line 27: in "head position" until a constructor is exposed. ...
  line 31: ... Performs one step of "congruence", which takes a term and produces
  line 92: ... closes one conv goal "trivially", by using reflexivity
  line 146: by focusing on particular subexpressions.</p><p>See <a href="https://lean-lang.org/theorem_proving_in_lean4/conv.html">https://lean-lang.org/theorem_proving_in_lean4/conv.html</a> for more details. ...

As you can see, the <url> in original line 146 is completely broken (https://leanprover-community.github.io/mathlib4_docs/Init/Conv.html#Lean.Parser.Tactic.Conv.conv), while this PR fixes it. This PR also introduces changes that " is replaced by ", similarly for <, > and &, but these changes in HTML text does not affect rendering.

[acmepjz]

Need to ensure that #126 does not reappear.

[acmepjz]

have a reference to the upstream issue in the source

Done.

[hargoniX]

Looks fine to me, let's see if some users find pages that are broken again^^