config.http test fails now on some machines

knot-resolver:postinstall+config / config.http               FAIL

The test still reliably passes on some machines, but it now reliably fails on some others. Based on my reproducibility testing, normal dependencies do not affect this (nor 5.x vs. 6.x), so it's most likely triggered by some kernel-side changes.

Assertion same() failed for test below (marked 'not ok'):
Expected: nil
Got: 'connect: Connection refused'
not ok 29 - AF_UNIX connect(): connect: Connection refused
# Trouble in /build/knot-resolver-5.7.4/modules/http/http.test.lua around line 114
----------------------------------- stderr -----------------------------------
flush: No such file or directory
stack traceback:
	[C]: in function 'error'
	...t2.1-compat53-0.14.4-1/share/lua/5.1/compat53/module.lua:173: in function 'assert'
	...uajit2.1-http-0.3-0/share/lua/5.1/http/h2_connection.lua:441: in function 'ping'
	/build/knot-resolver-5.7.4/modules/http/http.test.lua:115: in function </build/knot-resolver-5.7.4/modules/http/http.test.lua:103>
	[C]: in function 'xpcall'
	/build/knot-resolver-5.7.4/tests/config/test_utils.lua:5: in function 'runtest'
	/build/knot-resolver-5.7.4/build/../tests/config/test.cfg:42: in function </build/knot-resolver-5.7.4/build/../tests/config/test.cfg:35>
==============================================================================

---

Note: nowadays people probably shouldn't even use the http module anymore, so... we might simply just disable this test or something.