Fix flaky test: full_client::client::legacy_and_unstable_block_subscription_reconnect

[jsdw]

I think I've seen this test fail more than once. Can we make it more robust and remove the chance of it failing?

running 1 test
test full_client::client::legacy_and_unstable_block_subscription_reconnect has been running for over 60 seconds
test full_client::client::legacy_and_unstable_block_subscription_reconnect ... FAILED

failures:

failures:
    full_client::client::legacy_and_unstable_block_subscription_reconnect

test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 85 filtered out; finished in 360.01s


--- STDERR:              integration-tests full_client::client::legacy_and_unstable_block_subscription_reconnect ---
thread 'full_client::client::legacy_and_unstable_block_subscription_reconnect' panicked at testing/integration-tests/src/full_client/client/mod.rs:415:1:
Test timedout: Elapsed(())

[jsdw]

Also the test is called legacy_and_unstable_block_subscription_reconnect but only looks like it tests the chainhead backend to me?

Running this locally it always passes in 25-30s, so maybe CI is just a bit slow and it sometimes takes a little over the default 60s timeout. PErhaps bumping it to 120s would be the quick fix, but we should think about what we're testing and maybe have a separate test for legacy and chainhead backend reconnecting?