scheduled_task: Refactor trigger definitions / validation logic in canonicalize_and_validate

Logic in the canonicalize_and_validate method takes a number of unique code paths specific to individual trigger types. It may be more useful to define some Ruby hashes that contain a "schema", so that validation logic could be more normalized / handled generically.

So for instance, something like

# base ITrigger

trigger_defaults => {

  fields => {

    'schedule' => { type => String, values => 'daily, weekly, monthly, once', required => true },

    'start_time' => { type => String, pattern => '\d\d:\d\d', required => true },

    'start_date' => { type => String },

    'minutes_interval' => { type => Integer },

    'minutes_duration' => { type => Integer },

  }

}

daily_trigger => {

  type => 2,

  fields => {

    'every' => { type => Numeric, values => [1, 2] },

    # etc

  }

}

monthly_trigger => {

  type => 4,

  fields => {

    'which_occurrence' => { type => Array, values => ['first', 'second', 'third', 'fourth', 'last'] }

    # etc

  }

}

I won't model them all out, but that should set the gist of things.

A full breakdown of all allowed values is in
https://gist.github.com/Iristyle/58ff446bd3f60e68d52df3868c917a0e

This is also likely the best way to fix the bug where mutually exclusive trigger hash settings are currently ignored. For example, a schedule of 'once' will accept an every value of '1', even though its unused, and there a number of other such examples.