unexpected error writing points to database: too many open files.

I was trying to write the following CSV data to influx and got the following error:

```
Error: Failed to write data: unexpected error writing points to database: [shard 33] open /Users/rsavage/.influxdbv2/engine/data/42a2da2da2bbc070/autogen/33/index/5/MANIFEST: too many open files.
See 'influx write -h' for help
```

https://www.kaggle.com/nphantawee/pump-sensor-data

```
➜ influxdb git:(master) ✗ influx write --format csv -f ~/Downloads/sensor.csv -b apps --skipHeader --header "#constant measurement,water_pump" --header "x|ignore,timestamp|dateTime:2006-01-02 15:04:05,sensor_00|double,sensor_01|double,sensor_02|double,sensor_03|double,sensor_04|double,sensor_05|double,sensor_06|double,sensor_07|double,sensor_08|double,sensor_09|double,sensor_10|double,sensor_11|double,sensor_12|double,sensor_13|double,sensor_14|double,sensor_15|double,sensor_16|double,sensor_17|double,sensor_18|double,sensor_19|double,sensor_20|double,sensor_21|double,sensor_22|double,sensor_23|double,sensor_24|double,sensor_25|double,sensor_26|double,sensor_27|double,sensor_28|double,sensor_29|double,sensor_30|double,sensor_31|double,sensor_32|double,sensor_33|double,sensor_34|double,sensor_35|double,sensor_36|double,sensor_37|double,sensor_38|double,sensor_39|double,sensor_40|double,sensor_41|double,sensor_42|double,sensor_43|double,sensor_44|double,sensor_45|double,sensor_46|double,sensor_47|double,sensor_48|double,sensor_49|double,sensor_50|double,sensor_51|double,machine_status|string" --header "#constant tag,id,pumpy_mcpumperson"
Error: Failed to write data: unexpected error writing points to database: [shard 33] open /Users/rsavage/.influxdbv2/engine/data/42a2da2da2bbc070/autogen/33/index/5/MANIFEST: too many open files.
See 'influx write -h' for help
```

Start with the `influx write` command and the supplied CSV invocation, then reproduce the failure using the linked pump-sensor data. Investigate why this write reaches the `too many open files` error; done means the same import completes without that database error.

Written by the indexing model from the issue text.