Description of problem:
-----------------------
The fix for https://bugzilla.redhat.com/show_bug.cgi?id=1419186 was to get Chargeback reports to skip hourly rollups that are older than the VM itself.

But, it looks like hourly rollups that are older than the VM are still being considered on days when there are less than 24 hourly rollups.

I have attached a screen shot of the hourly rollup values for 'cu-24x7' VM and a Chargeback report showing the costs for this VM.

From the attached screen shot, here's the values for the hourly rollups 
on 3/27/2017:
45.4 MHz, 42.8 MHz, 42.8 MHz, 41.4 MHz, 44.6 MHz, 43.4 MHz, 45.8 MHz, 43.6 MHz

The 'CPU Used' reported in the Chargeback report is 58.31 MHz.

Based on the hourly rollup values, CPU Used 
          = (45.4 + 42.8 + 42.8 + 41.4 + 44.6 + 43.4 + 45.8 + 43.6) / 6
          = 58.3 MHz

Note that there are 8 hourly rollups, 2 of which are older than the VM itself.
But, we are summing up all the 8 values, instead of skipping the 2 hourly rollups that are older than the VM and then that sum is being divided by 6.

Since the resource usage reported is inaccurate, the Chargeback cost reported is also incorrect.

cu-24x7 creation timestamp 2017-03-27 17:02:51.380219

vmdb_production=# select timestamp, cpu_usagemhz_rate_average from metric_rollups where resource_name = 'cu-24x7' and capture_interval_name = 'hourly' and timestamp < '2017-03-28'\x\g\x
Expanded display is on.
-[ RECORD 1 ]-------------+--------------------
timestamp                 | 2017-03-27 17:00:00
cpu_usagemhz_rate_average | 42.7944444444444
-[ RECORD 2 ]-------------+--------------------
timestamp                 | 2017-03-27 16:00:00
cpu_usagemhz_rate_average | 45.425
-[ RECORD 3 ]-------------+--------------------
timestamp                 | 2017-03-27 18:00:00
cpu_usagemhz_rate_average | 42.8277777777778
-[ RECORD 4 ]-------------+--------------------
timestamp                 | 2017-03-27 19:00:00
cpu_usagemhz_rate_average | 41.3888888888889
-[ RECORD 5 ]-------------+--------------------
timestamp                 | 2017-03-27 20:00:00
cpu_usagemhz_rate_average | 44.6388888888889
-[ RECORD 6 ]-------------+--------------------
timestamp                 | 2017-03-27 21:00:00
cpu_usagemhz_rate_average | 43.4444444444444
-[ RECORD 7 ]-------------+--------------------
timestamp                 | 2017-03-27 22:00:00
cpu_usagemhz_rate_average | 45.8
-[ RECORD 8 ]-------------+--------------------
timestamp                 | 2017-03-27 23:00:00
cpu_usagemhz_rate_average | 43.5666666666667


Version-Release number of selected component (if applicable):
------------------------------------------------------------
5.8, 5.7


How reproducible:
----------------
Always


Steps to Reproduce:
-------------------
1.Manage a provider.Enable C&U collection for the provider.
2.Generate Chargeback report for any VM


Actual results:
--------------
Inaccurate resource usage and Chargeback costs reported for days where consumed hours < 24 since old hourly rollups are still being considered.

Based on the hourly rollup values, CPU Used 
          = (45.4 + 42.8 + 42.8 + 41.4 + 44.6 + 43.4 + 45.8 + 43.6) / 6
          = 58.3 MHz


Expected results:
-----------------
The resource usage and Chargeback costs for days where consumed hours < 24 should be accurate.Any hourly rollup older than the VM should be skipped.

CPU Used = (42.8 + 41.4 + 44.6 + 43.4 + 45.8 + 43.6) / 6
         = 43.6 MHz

Additional info:
----------------