Add SaveStartupConfig for SaveOnSet feature

[rlucus]

sonic-net/SONiC#1239

[tomek-US]

Approving based on Sachin's approval.

[kwangsuk]

A question -
Say, config changes are made from other apps, but 'config save' not issued yet on the device.
Meanwhile, enabling save-on-set does forcefully save all the changes including one of the last gnmi-set operation?

[rlucus]

Updated CL with agreed upon method of using minimum version of rpc_copy

[rlucus]

A question - Say, config changes are made from other apps, but 'config save' not issued yet on the device. Meanwhile, enabling save-on-set does forcefully save all the changes including one of the last gnmi-set operation?

Yes that would be the case.

[mbalachandar]

Its out-of-date with base branch, please merge the latest from sonic-net:master

[rlucus]

Done

[sachinholla]

Latest commit is introducing many rpcs and options that are not really supported. This is not a clean approach. Approach of having a translib API would have been better -- copy r s is a standard operation in NOS world.

[anand-kumar-subramanian]

Latest commit is introducing many rpcs and options that are not really supported. This is not a clean approach. Approach of having a translib API would have been better -- copy r s is a standard operation in NOS world.

Anand>> Please have only the necessary RPC implemented end to end. Others will come in when Dell plans to bring it in.

[rlucus]

Latest commit is introducing many rpcs and options that are not really supported. This is not a clean approach. Approach of having a translib API would have been better -- copy r s is a standard operation in NOS world.

Anand>> Please have only the necessary RPC implemented end to end. Others will come in when Dell plans to bring it in.

They are returning Unimplemented. Do you want them removed from the yang and annotation as well? Please be more specific, it seems as if Sachin is rejecting the whole concept of the compromise we all agreed on.

[anand-kumar-subramanian]

Latest commit is introducing many rpcs and options that are not really supported. This is not a clean approach. Approach of having a translib API would have been better -- copy r s is a standard operation in NOS world.

Anand>> Please have only the necessary RPC implemented end to end. Others will come in when Dell plans to bring it in.

They are returning Unimplemented. Do you want them removed from the yang and annotation as well? Please be more specific, it seems as if Sachin is rejecting the whole concept of the compromise we all agreed on.

Anand>>I meant only the RPC to perform config save needs to be implemented end to end all others will be implemented when Dell plans to bring it in. Other RPCs should not be in the YANG, annotation or the transformer implementation.

[rlucus]

Anand>>I meant only the RPC to perform config save needs to be implemented end to end all others will be implemented when Dell plans to bring it in. Other RPCs should not be in the YANG, annotation or the transformer implementation.

Removed the requested stubs. I'm going to have to come back to this in a few days to figure out why the linter doesn't like removing them.

[rlucus]

Closing this PR as its no longer needed due to using sonic services client in sonic-net/sonic-gnmi#108