Fixed problems in handshake handling

[ethouris]

Fixes:

• the rejection case in startConnect was catching a not possible behavior. The condition stayed, but it is turned into just an IPE error message.
• added case when the processing ended with CONN_REJECT and was not properly handled - this adds the exception info
• removed calls to removeConnector previously applied when the process gets broken due to rejection. Instead the removal is done only in case when this function is about to finish and throw exception due to any case of abnormal result.
• in processRendezvous the "response" packet is checked if it contains data (otherwise payload size is set to -1). There were cases when a non-received payload packet was passed to this function and was interpreted as infinite-size content causing a crash
• made the updateConnStatus respond anything only in asynchronous mode (this was probably allowed in synchronous mode in UDT because it was an accidental patch to a bug). Keeping this running in synchronous mode might in specific cases cause processing the incoming packet in two FAs and two threads separately in areas that weren't predicted to be protected against simultaneous access
• fixed a problem when setting REQ-TIME LOW when no packet has arrived
• passing the rst state down to processRendezvous so that it can distinguish after-packet call to periodic-update call
• processRendezvous when called for periodic update (and is unable to get the data from KMREQ), crafts KMRSP from data recorded when doing reverse-KMX (WARNING: this makes this PR now depedent on Applied fixes for crypto in RX-to-TX cloning #365)

[ethouris]

INFORMATION FOR Status: Blocked: Requires #365 to be merged first.