Currently, the tests for rest streaming depend on a single seed value configured globally within the filehere. This makes the tests non-deterministic and inter-dependent making it difficult to understand and debug test failures.

Instead, we can either:

• setrandom.seed explicitly in each test function. The value can either be hard-coded in (recommended), or it can be varying (by passing in, say, an integer based on time of day, PID, etc.) .

or

• pass in the seed to the constructor of_ResponseItr within the test file.

In either case, make sure the logs and / or errors print out what the value of the seed turned out to be, so that if we encounter the same failure, we can make sure the test function sets the same seed so we can replicate it.

Also do the same for the test cases of asynchronous rest streaming once implemented. See: [async rest]streaming(#686)

Reference:#686 (comment)