Fix zstd:flush/2

[rickard-green]

Fix zstd:flush/2 introduced in #10511 @essen @jhogberg

In debug mode, the zstd NIF reduce the chunk size to 5 which causes the zstd:flush/2 function in the cstream testcase to return {flush, _} which causes the testcase to fail. {flush, _} is not a valid return value according to the docs.

This PR prevents the testcase failure, but the change in this PR is a guess of what the flush/2 function should have done. I'm not sure if this fix is correct or complete.

[github-actions]

CT Test Results

    2 files    100 suites   1h 7m 15s ⏱️
2 288 tests 2 236 ✅ 52 💤 0 ❌
2 706 runs  2 650 ✅ 56 💤 0 ❌

Results for commit 8413ea4.

♻️ This comment has been updated with latest results.

To speed up review, make sure that you have read Contributing to Erlang/OTP and that all checks pass.

See the TESTING and DEVELOPMENT HowTo guides for details about how to run test locally.

Artifacts

• Complete CT logs (Download Logs)
• HTML Documentation (Download HTML Docs)
• No Windows Installer found

// Erlang/OTP Github Action Bot

[jhogberg]

This violates the spec too, the output has to be {continue, iovec(), binary()} or {continue, binary()}, we can't mix it up.

Suggested change

	{continue, [Output | Output2]};
	{continue, iolist_to_binary([Output, Output2])};

[essen]

What happens if that output is sent directly to a file/socket, isn't the iolist_to_binary unnecessary there? Perhaps zstd can be updated to allow it to return more than binary() output?