- make rcu wait when its buffer gets full, instead of crashing
- fix makefile to compute dependency info as a side-effect of compilation (-MF)
- investigate 16 byte CAS 
    - ht can store GUIDs inline instead of pointers to actual keys 
    - mem can keep tail pointers for free-lists and do O(1) appends
- test ht
- optimize tracing code, still too much overhead