The $8- is being handled differently also, because the space in the regex filter shifts the normal #channelname at $7 into being the $8 word.
[Spamfilter] AndroUser2!~androirc@ip.address.here.threembb.co.uk matches filter '(?-i)[A-Z ]{30}': [PRIVMSG #testchannel: 'THIS IS A TEST TRIP OF THE SPAMMERSFILTER I SET ON THE NETWORK FROM SLEEPINGDRAGOONY'] [caps or netspammer] at 22:05:51 on 30/05/2021 on network.here
$2 = AndroUser2!~androirc@ip.address.here.threembb.co.uk
$3-4 = matches filter
$5 'filter':
$6 = [PRIVMSG
$7 #testchannel:
$8- = 'THIS IS A TEST TRIP OF THE SPAMMERSFILTER I SET ON THE NETWORK FROM SLEEPINGDRAGOONY'] [caps or netspammer] at 22:05:51 on 30/05/2021 on network.here
Without the space in the filter, the $8- begins with the singlequote followed by what they said. With the space, the $8- contains the "#testchannel:" string.
Also, the way you're tokenizing by 93, your output is terminated prior to the "[caps" string, so it includes the ending ] brace preceding that. However, it also means that if their message contains the ] character, the message will truncate at that instead.
^'([^']+)': \[\S+ \S+: '(.+)'\] \[
With a regex like the above against $5- it should create $regml(1) containing the filter assuming it does not contain any singlequotes followed by $noqt(: [) and creates $regml(2) containing their sentence assuming the sentence doesn't contain: $noqt('] [)
