fixing issues with benchmark Tests

[kiran-bjn]

I noticed 3 issues in the benchmark example code.

• When running the benchmark on a centrifuge-server running remotely, Clients reconnecting due to network issues causes the program to crash due to improper waitGroup actions
• When Clients reconnect, they can lose messages and can end up waiting for the messages to be received, indefinitely
• Converting the generated byte array for the payload to string before marshalling, converts the empty 0 bytes into the string representation, which is actually 6 times larger than the actual intended msg size. eg: ( [0] becomes "\u0000" )
  ref : https://goplay.tools/snippet/tdq61T0axc_E

The fixes respectively are

• Call Done() on the waitGroup only on Client connects and not on reconnects by maintaining a connected flag
• Accept a deadline time as an argument and timeout the test if it is active beyond the deadline time and print the accumulated benchmark report
• Don't stringify the byte array before marshalling

[FZambia]

Many thanks for these awesome improvements @kiran-bjn