tsdb: Fix split16 to avoid indexing 14 docs from the end

[pquentin]

We drop 2 documents from the corpus so that the document count is a multiple of 16 which will allow Rally to split at exact boundaries.

To confirm the fix, I've run the following queries at 2% of indexing and nothing from 2021-04-29 got indexed:

% curl http://localhost:39200/tsdb/_search --data-binary @d.json -H 'Content-Type: application/json' | jq .                                                   
{                                                                                                                                                              
  "took": 2557,                                                                
  "timed_out": false,                                                          
  "_shards": {                                                                                                                                                 
    "total": 1,                                                                
    "successful": 1,                                                                                                                                           
    "skipped": 0,                                                                                                                                              
    "failed": 0                                                                                                                                                
  },                                                                                                                                                           
  "hits": {                                                                    
    "total": {                                                                 
      "value": 10000,                                                          
      "relation": "gte"                                                        
    },                                                                         
    "max_score": null,                                                         
    "hits": []                                                                                                                                                 
  },                                                                           
  "aggregations": {                                                            
    "NAME": {                                                                  
      "buckets": [                                                             
        {                                                                      
          "key_as_string": "2021-04-28T17:00:00.000Z",                         
          "key": 1619629200000,                                                
          "doc_count": 2099499                                                 
        },                                                                     
        {                                                                      
          "key_as_string": "2021-04-28T18:00:00.000Z",                         
          "key": 1619632800000,                                                
          "doc_count": 484385                                                  
        }                                                                      
      ]                                                                                                                                                        
    }                                                                          
  }                                                                                                                                                            
}

with d.json having the following contents:

{
  "size": 0,
  "aggs": {
    "NAME": {
      "date_histogram": {
        "field": "@timestamp",
        "calendar_interval": "hour"
      }
    }
  }
}

[dliappis]

LGTM

[dliappis]

these two sentences might be useful to add in the Rally docs.