v0.8.6, build 168

Added possibility to configure a default provider for chats
Added architecture for future agent usage
Added a first agent to read, analyze and extract text from Markdown data
Improved the readability of the settings.json file by using indentation and enum names instead of numbers
Improved assistant overview; assistants will now wrap to the next line if there are too many to fit on the row
Increased the default value for the live translation delay from 1,000 to 1,500 ms
Fixed random number generator usage to be thread-safe