L10n problem: Strings have hardcoded capitalization

Some strings of the software are being turned into capital letters through css or something similar. This practice makes them difficult to translate, as many languages:
- Do not have capital letters (Persian, Arabic, Hebrew... )
- Have different spelling rules depending if the word is in capital letters or not (Greek and others)
- Do not use capital letters in the same way than English

The string 'START VPN' for example is 'Start VPN' on the translation file: https://hosted.weblate.org/translate/guardianproject/orbot/en/?checksum=d66d525b66046615 . Same with 'Choose how to connect' https://hosted.weblate.org/translate/guardianproject/orbot/en/?checksum=6a3a41df28044b66

For a better translation, please do not hardcode capitalization rules.

If you want the string to appear in all caps, it is better to write the source string in all caps and then translators will see what to do in their language.

Start by searching the Android source for the cited strings “START VPN” and “Choose how to connect”, then inspect where capitalization is applied. Compare those source strings with their Weblate entries; done means capitalization is no longer imposed on translated strings and the affected UI displays each language’s intended casing.

Written by the indexing model from the issue text.