Use CamelCase in handlers for less memory, more speed

[jmondo]

gsubbing the handler pattern strings from snake case to camel case uses a lot of unnecessary memory. profiler coming in the comments.

[jmondo]

I added this test:

def test_one_thing_memory
  report = MemoryProfiler.report do
    time = Chronic.parse("2012-08-02T13:00:00")
    assert_equal Time.local(2012, 8, 2, 13), time
  end
  report.pretty_print(to_file: 'memory.log')
end

[jmondo]

memory-before.txt
memory-after.txt

handler.rb 72,736 -> 5,120
chronic/lib 167,181 -> 94,445

[jmondo]

For benchmarking I ran the test suite 5 times on master and this branch

master
Finished in 2.317107s, 71.6411 runs/s, 308.1429 assertions/s.
Finished in 2.267655s, 73.2034 runs/s, 314.8626 assertions/s.
Finished in 2.133667s, 77.8003 runs/s, 334.6351 assertions/s.
Finished in 2.184228s, 75.9994 runs/s, 326.8890 assertions/s.
Finished in 2.094341s, 79.2612 runs/s, 340.9187 assertions/s.
(average 2.1993s)

camelcase
Finished in 2.096705s, 79.1719 runs/s, 340.5344 assertions/s.
Finished in 1.995551s, 83.1850 runs/s, 357.7959 assertions/s.
Finished in 2.004112s, 82.8297 runs/s, 356.2675 assertions/s.
Finished in 2.016496s, 82.3210 runs/s, 354.0796 assertions/s.
Finished in 2.024613s, 81.9910 runs/s, 352.6600 assertions/s.
(average 2.0274s)

(-7.81%)

faster 👍

[jmondo]

let me know if this is too ugly. i have some other nicer looking but slightly less performant solutions. (still an improvement, though)