That's just it. Modbus is a well defined standard, but the HY is NOT talking Modbus, it is talking a misinterpreted Modbus. If you need to send read commands to write and must create a special interface to handle that then that is totally wrong and twisted and is NOT Modbus. That's why you could not make it work with UCCNC or Mach3. Both of those have Modbus interfaces but both assumes that the connected units use Modbus, not a twisted form of communication.
I don't know if it ever can be modified to become Modbus standard, but why would anyone do that when there is already a nice interface in UCCNC which supports Real Modbus? Using that interface reduces the macro complexity and the number of lines to very few, so why would this be an interesting solution to any REAL Modbus unit users? It is OK for the HY, and perhaps a few more, but any standard Modbus device is better handled by UCCNC.
I am not sure I understand what you mean. I know you spent a great amount of time on this, but what you mean by no difference?
Of course it is possible to modify the macro, I was just point it out that by sending the message only once you are taking a risk and remove a safety feature which is built in in Modbus by default. It is absolutely NOT safe to send a message only once and ASSUME that it is received without error in 100% of the cases. If you do that you need to read back the answer also and check for confirmation, that is the least you can do if you are concerned about the data.
Communication issues are not always connected to noise, so there is no warranty that you will get problems elsewhere just because you occasionally would miss a beat.. I mean, in serial communication it is enough if a single bit is changed and the checksum will be wrong so the whole message will be discarded by the VFD, assuming it checks the checksum.