Use valid format for InfluxDBClient.write_points and convert metric path to metric name plus tags

[TomTheBear]

In my tests, the InfluxDB handler didn't work because the JSON dict created by the handler wasn't accepted by python-influxdb. I re-checked the implementation in python-influxdb and the problem is definitely on at the handler side. This PR includes:

• Proper dict keys for python-influxdb
• Transform the metric path to metric name plus tags. This is probably not complete, currently only hostname and if available CPU are added to tags. The metric name will be _
• Remove the timestamp test for sending a batch. This could cause a loss of metrics when the batch buffer is almost full and the handler should process a bunch of metrics coming in in a small time range.

[coveralls]

Coverage decreased (-0.09%) to 24.578% when pulling a0160c0b484f56c560d3bc76e0e5c94f55f616bc on TomTheBear:InfluxDB_handler into f2bece462577a7c557be8a9f90f6b9340c3db571 on python-diamond:master.

[coveralls]

Coverage decreased (-0.09%) to 24.578% when pulling a0160c0b484f56c560d3bc76e0e5c94f55f616bc on TomTheBear:InfluxDB_handler into f2bece462577a7c557be8a9f90f6b9340c3db571 on python-diamond:master.

[TomTheBear]

There is already the PR #504 doing the same stuff