[Olly] [Salesforce] Fix retrying when Salesforce server sends EOF using cometD input

[milan-elastic]

Describe the enhancement:

Currently, Salesforce is sending unexpected EOF while open client long pulling, and CometD is not able to handle this unexpected issue. which we need to explicitly handle either from the library we are using or the comet input itself. Finalize an approach to handle the same.

For more information please check an existing issue from salesforce.

[kush-elastic]

Here's till now information that we got from the existing issue from CometD repo issues:Issue link

• Based on what is described in the above-linked issue, we are probably facing the same issue as the salesforce CometD setup is not able to explicitly handle EOF errors.
• Basically, CometD Should have already identified that the error is EOF, still it is returning unexpected EOF.
• Normally, we would receive io.EOF from Golang req.DO() if it expected an EOF error from the CometD handler. which seems to be already handled in the elastic/bayeux library that we are using.
• Based on a comment from one of the CometD repo handler, what we need to do in this kind of case is re-handshake and create a new client. To do that we need to identify the exact error and in only that case we can re-handshake.