fix(tests): do not specify port number

[LNSD]

This PR should help mitigate the "LPError: Address already in use" issues reported in #1357.

From this conversation at discord:

FWIW. The issue could be solved by allowing the Waku node/switch to be configured with a /ip4/0.0.0.0/tcp/0 (then it would look for a free TCP/UDP port and bind to it). Then each test case would use its own TCP port, avoiding the "port reuse".

[alrevuelta]

lgtm! thanks!

(ingnore this if CI is ok)
found in the past some cases where the supplied port was 0 and some tests were failing because the assert was (==0) but in the multiaddress what was being stored was a "real" available port (eg 2000) so 0!=2000.

[rymnc]

Thanks for this! LGTM if ci passes in the first go :)

[LNSD]

found in the past some cases where the supplied port was 0 and some tests were failing because the assert was (==0) but in the multiaddress what was being stored was a "real" available port (eg 2000) so 0!=2000.

I explicitly decided to not modify those test cases exactly for that reason. Those test cases will have to be fixed in a different piece of work.

[LNSD]

Passing ✅ 🎸