[test] bump timeout in feature_init for troubleshooting CI failure

Summary:
This test, to see if the node fails to start in case some files are missing, is not supposed to take more than one second, but on CI with the address sanitizer we sometimes (avbout 10% of the time) see that the bitcoind process still hasn't returned after 300s, even though all the expected lines are already in the log, including [Shutdown] Shutdown: done. It is as if the ASAN instrumentation prevents the process from disappearing from the process table.

I was not able to reproduce this issue locally, even with much shorter timeouts.

Let's double the timeout for troubleshooting, to see if timeouts affect the failure rate or if the process is just stuck forever no matter how long we wait.

Test Plan: ninja check-functional

Reviewers: #bitcoin_abc, Fabien

Reviewed By: #bitcoin_abc, Fabien

Differential Revision: https://reviews.bitcoinabc.org/D18800