Build options from per-point-type factories

[jadamcrain]

Item 7 of #59.

Factories

• PointOptions builds the options that several kinds of point repeat:
  • the events generated for the count-limited tests
  • the events' class
  • the default event variation
  • the points kept at only one width (only16Bit / only32Bit)
  • for outputs: the installed points, the uninstalled index, the points per request and the changeable status flags
• ClassOptions.EventClass builds each event class's options.
• TimeSyncOptions.TimedEvents builds the with-time and relative-time options of binary and double-bit inputs.
• BinaryOutputOptions builds one set of CROB fields, which the primary and alternate commands and the control code points share.

Renamed keys

These keys differed only by accident, and now follow one pattern:

Before	After
control.*	binaryOutput.*, like analogOutput.* and supports.binaryOutput.* (ControlOptions is now BinaryOutputOptions)
binaryInput.numBinaryEventsGenerated, doubleBitInput.numDoubleBitEventsGenerated, counter.numCounterEventsGenerated	<group>.numEventsGenerated
binaryInput.binaryEventsClass, doubleBitInput.doubleBitEventsClass, counter.counterEventsClass, analogInput.analogInputEventsClass	<group>.eventsClass
control.maxControlPointsInSingleRequest, analogOutput.maxAnalogOutputsInSingleRequest	binaryOutput.maxPointsInSingleRequest, analogOutput.maxPointsInSingleRequest
supports.doubleBit*, supports.analog* (e.g. supports.disable.analog, supports.analogChangeEvents)	supports.doubleBitInput*, supports.analogInput*, matching the input groups
supports.analogEventsQueuedPerPoint	supports.analogInputEventsQueuedPerPoint
event.buffer.maxDoubleBitBinaryInputEvents	event.buffer.maxDoubleBitInputEvents

A configuration file with an old key now gets an unknown-option error. stepfunc/dnp3's conformance configuration will be updated to match.

Unchanged

The options' defaults, bounds and allowed values are unchanged, and so is their order in the written configuration file.

I compared a before/after dump with the renamed keys mapped back to their old names. It covered:

• every option
• both revisions' trees
• every procedure's -describe output
• the disabled lists
• Revision.validate

It ran at each subset level, with a device profile, every boolean option on, and every boolean option off. The only differences are in option descriptions that now read the same for each kind:

• "binary output points" where it said "control points"
• "counter events" where it said "binary counter events"
• the frozen counters' width options, which now spell out what they mean