iceRestart on createoffer not working

Hello,

I'm implementing iceRestart when the connection of onIceStateChange() goes to disconnect/failed.
In our case this can happen, while on the phone, we switch from ethernet to wifi, or in general switch IP addresses.

So I need to implement iceRestart to re-negotiate the ice candidates.
I can force a new CreateOffer with iceRestart:true and I get new icecandidates.
at the end a new 'invite' is send to the websocket, notifying the webrtc signalling server of Amazon Connect.

The problem is that it is not reacting well to sending an 'invite' again on a session that is already ongoing.

I receive a 'BAD_REQUEST' from the signalling websocket :

{\"connectionId\":\"99b1dc8c-01c4-44aa-xxxxxxx\",\"jsonRpcMsg\":{\"id\":\"ea012284-7d85-4c14-969f-xxxxxx\",\"jsonrpc\":\"2.0\",\"error\":{\"code\":400,\"data\":{\"errorCause\":\"BAD_REQUEST\"},\"message\":\"Error\"}}}"

My questions, is it possible to have an iceRestart with this library, and am I missing something why it is not working?
For completeness the initial createOffer flow and and the restartIce Flow are exactly the same. instead of using an 'invite' should i use another method in this case so amazon connect signaling server does not react with a BAD_REQUEST?

thanks for the assistance,

Start by tracing the createOffer and restartIce flows described in the issue, then inspect how the signaling WebSocket handles the repeated invite on an ongoing session. Compare the initial negotiation with the restart sequence and determine the supported Amazon Connect signaling method; done means an ICE restart completes without the BAD_REQUEST response.

Written by the indexing model from the issue text.