Consider a narrower dictionary for message-id generation
Currently, mutt message-id generation uses the base64 dictionary for random string generation. While this perfectly conforms to the RFC, the base64 dictionary includes URL-unfriendly characters such as + and /. This complicates message-id based lookups with services like public-inbox (e.g. on lore.kernel.org and similar sites).
Please consider switching to a narrower dictionary such as base32 for message-id generation. This will have no negative impacts on mutt and will help avoid problems with message lookup and retrieval.