[opentitantool] Consider adding the CRC check back to update-usr-access command

[alphan]

14877 adds the update-usr-access to opentitantool and uses it in bitstream_splice to improve local workflows that involve splicing a locally built rom or test_rom into a bitstream from the GCS bucket. See #14865 for details.

One particular shortcoming of the implementation in #14877 is that it disables the CRC check during FPGA programming. We should consider enabling it again.

[dmcardle]

Taking a stab at sketching out the subtasks:

• [ ] Compute the CRC of the bitstream after updating its USR_ACCESS value.
• [ ] Overwrite the value following WRITE_CRC_REG with the new CRC, instead of overwriting both with NOOP.
• [ ] Handle the case that the incoming bitstream does not already have WRITE_CRC_REG? Or add an assertion?

[alphan]

Thanks @dmcardle! We should check the example bitstream that I quoted in the original issue but IIRC

• checksum starts at a RCRC instruction, computed in rolling fashion and written after each write crc command. So, we might not need to handle the no crc command case specially.
• they use crc32c.