Translating a previous output from `/translate` shouldn't translate username header from the original message

For example, if I send a message `"/translate hola"`, then another user calls `/translate` in response to the bot's output (bot's output would look like `"<Summit>\nHello"`), the text `"<Summit>\n"` (or `"<Summit via User2>\n"`) should be removed from the LLM input string.

I think the only way to fix this is to strip the first line if it regex-matches `<.+>\n`.