Heap dump scripts fail to on large clusters (200+ nodes)

[ddelnano]

The collect_heap_pprofs.sh script is our main tool for debugging PEM memory use. On large clusters, the results of the px._HeapGrowthStacks UDTF can be so large that the output is truncated. In order to work around this, we need asid filtering for that function so that large clusters can still have specific instances memory profiled.

An example of how this would work can be seen on this branch.

[ddelnano]

I forgot that the changes to the shell script need to be upstreamed still. Re-opening until that is done.