Fix the email check for untrusted comments

Your service pulls email addresses out of free-text comments that anyone can post. An assistant's first regex matched every example the team tried, but nobody could say what it accepted. The team asked for a check whose rules can be read and tested.

An assistant wrote is_plausible_email(candidate). It is in the editor. It has a bug. Find it and fix it.

Input and output:

• candidate: the text to check.
• Return true when it is a plausible email address, false otherwise.

Rules the tests check:

• There is exactly one @. It splits the text into the local part, before it, and the domain, after it.
• The local part is 1 to 64 characters. Each is an ASCII letter, a digit from 0 to 9, ., +, _ or -. It does not start or end with ., and never has two . in a row.
• The domain is 1 to 255 characters. Split on ., it has at least 2 parts, called labels.
• Each label is not empty, uses only ASCII letters, digits and -, and does not start or end with -.
• The last label is at least 2 characters long and is all letters.
• Any other character, a space included, makes the text not plausible.