[Failing Test][Go SDK]: Flake: TestElementChan/FillBufferThenAbortThenRead

### What happened?

--- FAIL: TestElementChan (0.00s)
11:55:05 --- FAIL: TestElementChan/FillBufferThenAbortThenRead (0.00s)
11:55:05 datamgr_test.go:400: got sum 21, count 21, want sum 20, count 20
11:55:05 FAIL

It's due to a concurrency race on handling the channel close from the side. The behaviour shouldn't cause a correctness issue since it's so we can clean up residual data after an instruction is terminated. The test is just too specific and should be relaxed a little.

### Issue Failure

Failure: Test is flaky

### Issue Priority

Priority: 2 (backlog / disabled test but we think the product is healthy)

### Issue Components

- [ ] Component: Python SDK
- [ ] Component: Java SDK
- [X] Component: Go SDK
- [ ] Component: Typescript SDK
- [ ] Component: IO connector
- [ ] Component: Beam examples
- [ ] Component: Beam playground
- [ ] Component: Beam katas
- [ ] Component: Website
- [ ] Component: Spark Runner
- [ ] Component: Flink Runner
- [ ] Component: Samza Runner
- [ ] Component: Twister2 Runner
- [ ] Component: Hazelcast Jet Runner
- [ ] Component: Google Cloud Dataflow Runner