Merge: lib/core: Optimized `html_escape` for FlatText variants

Merge: lib/core: Optimized `html_escape` for FlatText variants

As @privat requested, a byte-oriented optimized version of `FlatText` for `FlatString` as it is the most common case, like `escape_to_c`, it works at the byte level to accelerate fetching of characters.

On a short test program

~~~nit
var s = ""
for i in [0 .. 2000[ do s = "<STRING/&rt;".html_escape
~~~

Valgrind reports an old runtime of 18.208 MIr with a 8856 Ir/call to `html_escape`; the new runtime is 3.093 MIr, which translates to a 1298 Ir/call.

Pull-Request: #1775
Reviewed-by: Jean Privat <jean@pryen.org>
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>