Before this PR, the caching and computation of `IRBlock` spanned three
cache stages and was also separate from `SSAConstruction` even though it
shared some computations with it. They are now all cached together, so
the number of stages is reduced by 2 for each layer of IR.
I made the choice of what to cache be similar to what we do for
`PrimitiveBasicBlock` as I've recently benchmarked this and found it to
be a good choice.
The check that an instruction is in the same function as its operands is
hopefully redundant and can be removed. Just to be sure, I've added the
check to a sanity query.
This check turned out to cause bad performance in the alias analysis
because it got inlined into `AliasAnalysis::resultEscapes` and then
pulled out to a loop-invariant predicate that got a bad join order. With
this check removed, the `ssa/AliasAnalysis.qll` file is orders of
magnitude faster.
As reported in CPP-236, this query has false positives on signed
integers that cannot be negative. It could possibly be improved with a
local range analysis, but the query would most likely still have so many
false positives that we would have to lower its precision.
Under our current policy, this change will make the query hidden by
default on LGTM.
