Tidy up, and add harness for examples

[gerph]

This change adds a harness for the examples so that they can be run as part of automation, or to check what the performance is in a simple way. The changes are:

• Rename the csv example to zipfile, as that's what it tests
• Update the examples so that they can run on Python 2 (when the rest of the library is capable of doing so).
• Create a harness that allows the examples to be run through the fuzzer. This harness reports on what it found from the results, which can be useful in particular to report on the speed of all of the examples if performance is a concern.

The harness output looks like this:

charles@laputa ~/external/pythonfuzz (add-harness-for-examples)> python3 examples/run_all_examples.py
Looking in examples for examples
Example: codeop
  Executions       : 1000
  Corpus           : 7
  Coverage         : 360
  Final speed      : 653/s
  Memory           : 36.47 MB
  Runtime          : 1.99 s
  Overall speed    : 502.66/s
  Return code      : 0
Example: zipfile
  Executions       : 1000
  Corpus           : 2
  Coverage         : 106
  Final speed      : 452/s
  Memory           : 34.92 MB
  Runtime          : 2.10 s
  Overall speed    : 476.58/s
  Return code      : 0
Example: htmlparser
  Executions       : 1000
  Corpus           : 7
  Coverage         : 442
  Final speed      : 695/s
  Memory           : 35.75 MB
  Runtime          : 1.77 s
  Overall speed    : 566.00/s
  Return code      : 0
Example: zlib
  Executions       : 1000
  Corpus           : 2
  Coverage         : 35
  Final speed      : 385/s
  Memory           : 34.83 MB
  Runtime          : 1.33 s
  Overall speed    : 752.01/s
  Return code      : 0
Example: xml
  Executions       : 1000
  Corpus           : 4
  Coverage         : 69
  Final speed      : 852/s
  Memory           : 35.21 MB
  Runtime          : 1.47 s
  Overall speed    : 679.20/s
  Return code      : 0
Example: fuzzit
  Executions       : 1000
  Corpus           : 4
  Coverage         : 63
  Final speed      : 1075/s
  Memory           : 34.86 MB
  Runtime          : 1.22 s
  Overall speed    : 821.51/s
  Return code      : 0
Example: aifc
  Executions       : 1000
  Corpus           : 4
  Coverage         : 101
  Final speed      : 523/s
  Memory           : 35.07 MB
  Runtime          : 1.81 s
  Overall speed    : 553.42/s
  Return code      : 0

The number of runs by default is 1000, which is poor for testing performance, but command line option can change that as needed by the user.