Simplify max nicklen checks for hostmask parsing

[jwheare]

This removes checking isupport NICKLEN for isHostmaskNicknameOn (used for sender/prefix parsing) in all cases.

Reasons:

• This was explicitly disabled in several cases anyway (missing isupport entirely, znc)
• This would fallback to a default of 9 when no NICKLEN was provided in isupport, but a default of 50 when explicitly disabled
• NICKLEN is not strictly the maximum length of a nickname on an IRC server. See:
  ** https://tools.ietf.org/html/draft-hardy-irc-isupport-00#section-4.14 "A client elsewhere on the network MAY use a nick length of higher value"
  ** https://tools.ietf.org/html/draft-brocklesby-irc-isupport-03#section-3.13 " This parameter does not restrict the length of any nicknames other clients on the network may use"
  ** https://defs.ircdocs.horse/defs/isupport.html#nicklen "Other clients on the network may have nicknames longer than this."

The most serious issue here is for servers with ISUPPORT but no NICKLEN key, with nicks longer than 9 chars, where parsing just silently breaks and stuff like server-initiated JOINs fail to register.

This now always uses a limit of 50, which is arbitrary but probably a safe enough sanity check.