fix(ci): Fix unit flakes from testnode by stopping calling the stop functions multiple times

[evan-forbes]

trying a hack in CI

[evan-forbes]

I don't think I'm missing other failures, but this one is still present. afaiu this is not testnode related.

--- FAIL: TestRetriever_Retrieve (465.29s)
    --- FAIL: TestRetriever_Retrieve/128x128(max) (464.22s)
        retriever_test.go:61: 
            	Error Trace:	/Users/runner/work/celestia-node/celestia-node/share/eds/retriever_test.go:61
            	Error:      	Received unexpected error:
            	            	context deadline exceeded
            	Test:       	TestRetriever_Retrieve/128x128(max)
--- FAIL: TestFraudProofValidation (20.22s)
    --- FAIL: TestFraudProofValidation/ods_size:128 (8.39s)
        retriever_test.go:127: 
            	Error Trace:	/Users/runner/work/celestia-node/celestia-node/share/eds/retriever_test.go:127
            	            				/Users/runner/work/celestia-node/celestia-node/share/eds/retriever_test.go:108
            	Error:      	Received unexpected error:
            	            	before batch commit: context deadline exceeded
            	Test:       	TestFraudProofValidation/ods_size:128
FAIL
coverage: 75.3% of statements
FAIL	github.com/celestiaorg/celestia-node/share/eds	524.601s

[Wondertan]

FYI, we agreed internally to skip the retriever tests and have yet to make the change. These are old reconstruction tests, and this code is not used, so feel free to ignore it. As to why it's failing after being stable for a long time, I believe the flakes are coming from IPFS-related dependencies, since that is the only thing that changed in this code.

[renaynay]

Thank you for this @evan-forbes

Ran this locally and didn't hit any flakes. CI looks like only failing on retriever tests which are not resp of core-app 👍🏻

[walldiss]

Thank you!