tests: Refactor driver error assertions

[bkeryan]

What does this Pull Request accomplish?

Add new assertion macros and helper functions to test_helpers.h:

• EXPECT_THROW_DRIVER_ERROR
• EXPECT_THROW_DRIVER_ERROR_WITH_SUBSTR
• EXPECT_DRIVER_ERROR
  • Replaces expect_driver_error helper function
• EXPECT_DRIVER_ERROR_WITH_SUBSTR
• get_driver_error():
  • Accepts both gRPC metadata multimap types and grpc_driver_error exception
  • Returns int.
  • If the ni-error key is not found, get_driver_error() fails the test and returns 0.

Why should this Pull Request be merged?

Reduce boilerplate and increase consistency in tests.

What testing has been done?

Built on Windows and Linux.
Ran DAQmx tests on Windows.

[astarche]

Looks really good!

[bkeryan]

FYI, EXPECT_THAT(x, Optional(y)) prints the inner value even when gtest's PrintTo(const std::optional<T>&) overload is not available, but it also prints the std::optional as a bunch of bytes:

[ RUN      ] NiDAQmxDriverApiTests.AOVoltageChannel_WriteAODataWithOutOfRangeValue_ReturnsInvalidAODataError
C:\Dev\grpc-device\source\tests\system\nidaqmx_driver_api_tests.cpp(1170): error: Value of: driver_error
Expected: value is equal to -200560
  Actual: 8-byte object <8F-F0 FC-FF 01-CC CC-CC>, whose value -200561 doesn't match
[  FAILED  ] NiDAQmxDriverApiTests.AOVoltageChannel_WriteAODataWithOutOfRangeValue_ReturnsInvalidAODataError (19 ms)

[SparkingSpork]

Declining to review since I'm not code owner on any of the files now.

[bkeryan]

This broke the build because I added new calls to this function in #928. Fixed in #939