Stale index metrics

[Jiaweihu08]

The current version of the index metrics is not suited for datasets that have appends or have experiences optimization/replications.

Only cube level metrics are shwon and block of files level statistics are absent:

// EXAMPLE OUTPUT
OTree Index Metrics:
dimensionCount: 2
elementCount: 2879966589
depth: 9
cubeCount: 13141
desiredCubeSize: 500000
indexingColumns: ss_sold_date_sk,ss_item_sk
avgFanout: 4.0
depthOnBalance: 1.3567716601745503

Stats on cube sizes:
Quartiles:
- min: 456367
- 1stQ: 498510
- 2ndQ: 499954
- 3rdQ: 501410
- max: 536430
Stats:
- count: 3285
- l1_dev: 0.00449603896499239
- l2_dev: 1.3487574366807247E-4
Level-wise stats:
level, avgCubeSize, stdCubeSize, cubeCount, avgWeight:
- 0:	497810,		0,		1,	1.7361319627929786E-4
- 1:	494798,		3550,		4,	8.689350799817908E-4
- 2:	499781,		3488,		16,	0.003668841950401859
- 3:	500516,		4292,		64,	0.015534089088738918
- 4:	500289,		3967,		256,	0.06698862054431544
- 5:	499966,		3530,		1024,	0.287867372830027
- 6:	499962,		3040,		1792,	0.6729941083529944
- 7:	500142,		10508,		128,	0.7959112180321912

We need more statistics to describe the index with more than a single write.

Candidate metrics:

• Total file/block count
• Average file per cube
• Average file size
• Average records per file