Indexing fails due to missing locale

Problem/Motivation

The Tika extractor sets the locale to en_US.UTF-8 to avoid stripping UTF-8 multibyte characters. This probably works for most systems, however if that locale is not present the indexing fails - the extract function throws an exception Tika Exctractor is not available (which has a typo btw). AFAIK the C.UTF-8 locale is more likely to be present on a system, so would be a more reliable choice.

I've set the priority to Minor since I presume this is an unusual scenario: I'm using Lando on WSL (Windows Subsystem for Linux on Win10) with the Pantheon recipe. The only locales present on the container Lando builds are:

$ locale -a
C
C.UTF-8
POSIX

Steps to reproduce

1. Use a bare minimum Linux install, or remove the en_US.UTF-8 locale.
2. Create a file with UTF-8 characters in the filename, eg 'Drupal’s contributors.docx' (character 0x2019 apostrophe)
3. Index using the Tika extractor

Proposed resolution

Change line 39 and 56 of /src/Plugin/search_api_attachments/TikaExtractor.php to use the C.UTF-8 locale:

    // setlocale(LC_CTYPE, 'en_US.UTF-8');
    setlocale(LC_CTYPE, 'C.UTF-8');

    // shell_exec("LANG=en_US.utf-8");
    shell_exec("LANG=c.utf-8");