RFR(S): 8240676: Meet not symmetric failure when running lucene on jdk8

[Roland Westrelin]

Hi Tobias,

> webrev.01 looks good to me but have you thought about using something like a verification depth
> counter that bails out after a certain recursion depth instead of the flag?

What's your concern? That the patch is too intrusive? Or that it still
has too much overhead?

Another solution would be to attach the flag (or recursion depth) to
Compile. Patch would then be much saller but not sure polluting Compile
with this is better.

Roland.