Use 16-bit integer I/O rather than floating-point

[marktsuchida]

Currently we use DAQmxWriteAnalogF64() and DAQmxReadAnalogF64() for scan waveform output and detector signal input. This should be changed to use DAQmxWriteBinaryU16() and DAQmxReadBinaryU16().

For scan waveform generation, we want control over exact DAC units so that we can generate the linear trace with equal (voltage) spacing.

For signal input, we want to use the raw ADC units to produce our 16-bit image, because applying scaling (voltage calibration) can result in histogram artifacts.

The 16-bit I/O functions bypass the NI DAQ voltage calibration, but we do not benefit from exact voltage measurements for either input or output.

[marktsuchida]

We should also support selection of the input voltage range used by the DAQ (6356 provides the choice of ±1, ±2, ±5, and ±10V; currently we always use ±10V).
Closely related to #16.

Confusingly, we currently have a setting, Input Voltage Range, which does not set the DAQ input range, but does affect the sample scaling. Currently, pixel value = 65535 * (V + offset) / range, clamped to [0, 65535], where V is the voltage measured by the DAQ (so 1 count in the produced samples represents 1.525e-4 V, i.e., 6553.5 counts/V, under the default "10V" range; we map [-1V, +9V] to [0, 65535]).

For compatibility, perhaps we can have a SampleMode setting, with options Voltage (Floating Point) and ADC Counts (Integer).

• Repurpose Input Voltage Range. Instead of the current Float64, we should use Enum type with options like -1.0 to 1.0V, ..., -10.0 to 10.0V. This should default to the widest range supported by the hardware and should actually set the hardware range. But no longer use this for sample scaling.

• Add a DigitalGain setting.

  • When in voltage mode, this replaces the old 65535 / range (which in practice was always 6553.5; that can be the default gain).
  • When in ADC Counts mode, DigitalGain should only allow the value 1.0 for now (we may add 0.5, 2.0, etc., if we implement sample summing).
  • We could allow negative values (in case the detector pre-amp is non-inverting).

• Add an Offset setting.

  • The first question is whether to apply offset before or after gain.
    • If applied before, the offset is in meaningful units (volts, ADC counts) representing the max negative input we can capture.
    • If applied after, the offset is in pixel value units.
    • I think it makes more sense to apply before gain.