Make `array_size_in_samps` / `array_size_in_bytes` optional for DAQmx Read RPCs

[bkeryan]

array_size_in_(samps|bytes) is an artifact of the DAQmx C API's memory allocation strategy and does not exist in any other DAQmx API implementation (LabVIEW, .NET, Python). If this parameter was optional and grpc-device calculated the required buffer size automatically, it could simplify some use cases and/or prevent the need to make additional RPCs to query the number of channels, default samples to read, etc.

Here's how that could work for ReadAnalogF64:

• If you specify array_size_in_samps, then grpc-device passes that to the C API and everything works like today.
• If you don't specify array_size_in_samps, then grpc-device calculates the required array size based on num_samps_per_chan and ReadNumChans. If num_samps_per_chan == -1, then it also uses DefaultNumberOfSamplesToRead.

This also applies to ReadDigitalLines, ReadDigitalU32, ReadBinaryI16, ReadRaw, etc. Note that some of these require additional information besides ReadNumChans and DefaultNumberOfSamplesToRead:

• ReadDigitalLines needs ReadDigitalLinesBytesPerChan. ReadDigitalWaveforms already uses this info.
• ReadRaw needs ReadRawDataWidthInBits

See #1223 (comment) for more details.

AB#3465613