#42: Zoltan: Assess MPI_COMM_WORLD usage and implement a getter

[tlamonthezie]

Fixes #42

This is still a work in progress.

This PR includes:

• New function named MPI_Comm_Default() to return currently MPI_COMM_WORLD
• Call of MPI_Comm_Default() instead of MPI_COMM_WORLD in src/zoltan/src/driver, src/zoltan/src/fdriver, src/zoltan/src/util/memory_usage, src/zoltan/src/util/network_topology, src/zoltan/src/Utilities/Timer, src/zoltan/src/Utilities/Communication, src/zoltan/src/Utilities/Memory, src/zoltan/src/ch, src/zoltan/src/hier, src/zoltan/src/lb, src/zoltan/src/order, src/zoltan/src/phg, src/zoltan/src/tpls, src/zoltan/src/zz and developer README.

[thearusable]

@JacobDomagala @stmcgovern Zoltan contains also a directory called siMPI (src/zoltan/siMPI) which contains many MPI related source files and they look like a copy of MPI library. There are a various MPI related functions and MPI_COMM_WORLD is used in couple of places. Should this PR also include the changes in siMPI or that is something that should be left as it is?

[stmcgovern]

@JacobDomagala @stmcgovern Zoltan contains also a directory called siMPI (src/zoltan/siMPI) which contains many MPI related source files and they look like a copy of MPI library. There are a various MPI related functions and MPI_COMM_WORLD is used in couple of places. Should this PR also include the changes in siMPI or that is something that should be left as it is?

Thanks for picking this up @thearusable . In my opinion, I think that we should not touch siMPI.

[thearusable]

@JacobDomagala The global comm header is installed properly