Enable head-based sampling and add some scripts for large scale deployment

This PR is related to the following issues:

• Head based sampling: https://gitlab.mpi-sws.org/cld/tracing/hindsight-grpc/-/issues/17
• Data type of duration: https://gitlab.mpi-sws.org/cld/tracing/hindsight-grpc/-/issues/20
• Docker images and scripts for deployment: https://gitlab.mpi-sws.org/cld/tracing/hindsight-grpc/-/issues/15

Side note, gen_address.py and config/alibaba_parse.py are not used, will be discarded later.

README.md

@@ -293,6 +293,15 @@ completion.  Takes as argument the collector data file
       --usage                Give a short usage message
 ```
 
+### Generate addresses file (for Alibaba trace)
+
+To generate addresses file for deployment, first manually update the server hostname list in `gen_address.py`, Then run:
+
+```
+python3 gen_address.py -t config/alibaba_long
+```
+
+This would generate Hindsight config file for each service under `hindsight_conf/` and the addresses file under `config/`. Services are evenly deployed on all servers. Will need to copy `hindsight_conf/` to `/etc/` (See details in https://gitlab.mpi-sws.org/cld/tracing/hindsight/-/blob/main/docs/configuration.md).
 
 ### Explanation of topology file