Remove _fix_cid_text

[heijul]

The problem is not fixable. See the pdfminer.six faq. Chars that can not be read should only be used as delimiters for things like #46 or similar.

[heijul]

The current approach actually works for some cid codes, namely those for the german umlauts.

[heijul]

The current approach actually works for some cid codes, namely those for the german umlauts.

I doubt we can assume this in general.

Instead, we could do the following:

• Store the PDF BBox and the page of each cid char
• use the original, not preprocessed, PDF to extract the glyph (image)
• use a OCR library to detect the char from the image
• in case that fails, give up.