As a rule of thumb, if there're tests, they shouldn't be ignored: they help prevent changes that will introduce errors (or behaviour differences).
If you're using code from the master branch, it's very difficult to have a test failing because of a code change (but not impossible). My suggestion to skip the tests was only so that you could have a functional version...
There're a few tests that are "very sensible" on how much resources they can use - It's not uncommon having tests (momentarily) failing from time outs, memory exhaustion or other sporadic problem...
As I said on my former reply, you can look into the result files to check why the test failed. If you find an error, you may report it!
Oh, do you use the "-U" (or "update-snapshots") option? If not, you may be using some wrong/old dependency...
Regards,
Sérgio Ribeiro
Porto - Portugal.