Make the <~> operator equivalent to channel.read/channel.write

[dlongnecke-cray]

This PR makes the <~> IO operator throw, and insofar as possible equivalent in semantics to channel.read and channel.write. It also adds the throws keyword to dsi IO routines for domains and arrays, as a consequence of the above.

If a SystemError is thrown from within IO routines up through readThis/writeThis, then the error is caught and the equivalent QIO error code is stored in the channel. Otherwise, the error is unpacked and the EIO error code is stored in the channel.

As of this time there is no way to distinguish between SystemError thrown by the user and SystemError thrown by the Chapel IO machinery.

The test test/classes/deitz/class/tree.chpl was rewritten to use an iterative postorder traversal instead of a recursive iterator due to compilation bugs (likely due to #7134).

---

Testing:

• ALL on linux64 when CHPL_COMM=none and CHPL_LLVM=llvm
• ALL on linux64 when CHPL_COMM=gasnet and CHPL_LLVM=llvm

[dlongnecke-cray]

FYI, I plan to figure out why the test classes/deitz/classes/tree is segfaulting sometime early this week.

[mppf]

Generally this looks good to me.

The PR message has some out of date stuff (removing inline, e.g.). and says "Otherwise, the error is unpacked" but I think you mean "Otherwise EIO is stored in the channel".

I'll discuss your open questions with you separately.