Automatic detection of encoding not used for JS, JSON (and CSS) files

[benoit74]

Currently, we have a custom detection of file encoding, but unfortunately this is used only for HTML content.

This is not used for JS and JSON content, making some of these rewrites fail.

For CSS, we let tinycss use its own logic to detect encoding.

We should use the same automatic detection of encoding for all files (including CSS for rationality).

[benoit74]

This bug has caused the failure of:

• https://farm.openzim.org/pipeline/21cfa1f8-5397-4699-86c2-e8e0b7947808
• https://farm.openzim.org/pipeline/b113b7ea-b5a6-4012-acf8-a28150393351