Check bloom filter in Get function

[rjl493456442]

This pull request enhances the 'Get' function by incorporating a bloom filter check before loading data. This offers the advantage of minimizing avoidable disk reads, thereby contributing positively to the overall read performance.

The implementation is quite straightforward, as recommended by @jbowens. I use iterator.SeekPrefixGE, a method that internally performs a bloom filter check.

Informs #197

[cockroach-teamcity]

This change is 

[rjl493456442]

-- random-read-filter-exp-1000mb-non-existent-cp (2046 events) total time: 278.4032s
 total size: 1047756600 bytes
  mean mb/s: 3.589 (+- 0.707)
-- random-read-filter-master-1000mb-non-existent-cp (2046 events) total time: 368.5611s
 total size: 1047756600 bytes
  mean mb/s: 2.711 (+- 0.405)

Construct a 1.8gb size database and perform reading for non-existent entries, this PR is slightly faster than master.

[sumeerbhola]

Thanks for the contribution.

Reviewed 3 of 3 files at r1, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @rjl493456442)

---

level_iter.go line 811 at r1 (raw file):

	if l.split != nil {
		n = l.split(l.iterFile.LargestPointKey.UserKey)
	}

can you add
// Else l.split == nil, so we use the length of the key. This case can occur when getIter uses SeekPrefixGE.

[rjl493456442]

I have added the comments, please take another look.

[sumeerbhola]

Looks good!