[test] Remove zzz prefix from high memory browser tests

[sbc100]

• remove unneeded require_wasm64 hack (the default node version we
  ship now supports wasm64 so it is no longer needed)
• remove unneeded requires_v8 from high memory tests
• remove unneeded @no_wasm64 from test_4gb_fail

[kripken]

The zzzs were to defer allocating huge amounts of memory to the end, so any OOM there wouldn't affect anything else. I guess these tests have been passing stably for a while now, so it's ok to change that ordering?

[sbc100]

The zzzs were to defer allocating huge amounts of memory to the end, so any OOM there wouldn't affect anything else. I guess these tests have been passing stably for a while now, so it's ok to change that ordering?

I think so yes.. are you saying that once a large allocation fails a small allocation might also fail? that doesn't make a lot of sense to me.

[kripken]

are you saying that once a large allocation fails a small allocation might also fail?

Yes, exactly. If the browser tab OOMs - "Aw, snap" page - then our test harness is dead in the water and everything after it is lost. At least at some point in the past I saw failed allocations lead to such OOMs, though in principle you'd hope they'd be JS exceptions instead. A long time ago someone explained to me that in Firefox it's very hard to make them all JS exceptions for technical reasons, since a failed allocation could happen in so many different places.