Remove ledger purge batching

[sakridge]

Problem

Batching purges seems to lower performance overall since each delete_range seems to be constant time. Nodes can get into a situation of deleting a huge range of slots and they will issue many delete_ranges which each cause writes to the db and slowing down insert.

Some example of the node in tds that died #10821:

[2020-06-26T06:58:10.443246600Z INFO  solana_core::ledger_cleanup_service] first_slot=0 total_slots=386755 total_shreds=50039179 max_ledger_shreds=50000000, iterate_time took 323ms
[2020-06-26T06:58:10.443759936Z INFO  solana_core::ledger_cleanup_service] purging data from slots 0 to 20606000

It had 50m shreds in 386k slots. It probably only had data back to slot: 20,606,000 - 386,755 = 20,219,245. But it walked from slot 0 all the way up.

Summary of Changes

Don't purge in batches, just do one delete_range.

Fixes #

[codecov]

Codecov Report

Merging #10830 into master will decrease coverage by 0.0%.
The diff coverage is 92.3%.

@@            Coverage Diff            @@
##           master   #10830     +/-   ##
=========================================
- Coverage    81.8%    81.8%   -0.1%     
=========================================
  Files         303      303             
  Lines       71087    71067     -20     
=========================================
- Hits        58220    58192     -28     
- Misses      12867    12875      +8