Improve query efficiency: minimise data processed per query request.

[time-trader]

Currently any sensor data query processes the whole aerosense-twined.greta.sensor_data table.
SELECT * Here Instead of SELECT datetime, "f0_", "f1_", ..... etc.

Problem

Currently the table contains around 50GB of data. So 20 Queries will result in 1TB of data processed costing around 5 USD at current pricing.

Solutions

The query is charged according to the total data processed in the columns we select, and the total data per column is calculated based on the types of data in the column. So:

1. Immediate solution: select only the datetime and sensor_value columns.
2. Longer term solutions to discuss:
   2.1 Create permanent tables for each sensor_type with columns for each senor
   2.2 Introduce table partitioning (for example on daily basis)

[cortadocodes]

https://cloud.google.com/bigquery/pricing#on_demand_pricing

[time-trader]

And for the best practice:
https://cloud.google.com/bigquery/docs/best-practices-costs
and
https://cloud.google.com/bigquery/docs/best-practices-performance-communication