RFR (S) - 8249663: LogCompilation cannot process log from o.r.scala.dotty.JmhDotty

[eric.caspole at oracle.com]

Hi everyone,
Could I get reviews on this bug fix to LogCompilation tool. There were 
actually 2 problems, the first is that "site" in LogParser is not reset 
after a <parse>, and so the next parse could misuse the stale site which 
led to the stack trace in the bug.

The second problem appeared after I put the first fix in place, that 
<replace_string_concat> can have an uncommon trap 
reason='intrinsic_or_type_checked_inlining' so we need to record the 
location of that to process the uncommon trap that will follow it.

I tested this with logs from some jvm08's and a short run of all 
renaissance, and about 80 runs of the specific renaissance dotty that 
usually showed the first problem.

JBS:
https://bugs.openjdk.java.net/browse/JDK-8249663

webrev:
http://cr.openjdk.java.net/~ecaspole/JDK-8249663/01/webrev/

Thanks,
Eric