Bug 1089104

Description of problem:
This is a newly introduced problem now that we just changed the alogrithm to pick the smallest value in the graph data and and use its scale. For instance, if we have values of MB and GB, it would choose MB as the base scale and represent everything in MBs.

This is mainly a usability design issue in how we graph extreme ranges. If we have say less than 5 values that are in TB and the rest are in MB should those spike bars be shown as they skew the graph and we lose detail in the majority of the values. 
Possible solution: Perhaps we should truncate those spiky values at the top of the graph and put a symbol indicating out of bounds and hovering over this symbol would provide the detail of the symbol.


Version-Release number of selected component (if applicable):
RHQ 4.11 (and JON 3.2.2)


How reproducible: only with values that have a range of a couple scales (1,000,000x) - each scale represents 1000x and it requires more than 2 scales to see this problem - values that have hundreds and values in millions.



Steps to Reproduce:
You would need a huge dataset with a high degree of variance. Larry has provided a script to produce this in BZ1080470.

Actual results:
The YAxis values become so large that they begin to get clipped off to the left of the screen. This is due to values being represented in the minimum value's scale (say Bytes and remember there are values of Bytes in the data) and there are extreme values (say TB) a value of 1,000,000,000,000,000 on the graph


Expected results:
1) YAxis scale should show all values without being clipped.
2) Spikes overwhelm the small detail data and it gets lost (even though that may be where a majority of the data is clustered).


Additional info: