Batching POST request is locked on nanosecond timeprecision

__Steps to reproduce:__

Instantiate the session similarly to:
```
influxdb = InfluxDB2::Client.new("http://#{INFLUXDB_HOST}:8086",
INFLUXDB_TOKEN,
bucket: INFLUXDB_DATA, org: INFLUXDB_ORG,
use_ssl: false,
precision: InfluxDB2::WritePrecision::SECOND)
write_options = InfluxDB2::WriteOptions.new(write_type: InfluxDB2::WriteType::BATCHING,
batch_size: 100, flush_interval: 90_000,
max_retries: 3, max_retry_delay: 15_000,
exponential_base: 2)

influx_write_api = influxdb.create_write_api(write_options: write_options)

influx_write_api.write(data: )
```

__Expected behavior:__
The data should be sent with seconds time precision.

__Actual behavior:__
The data is sent with nanoseconds time precision, and is not recognized by Influx since its timestamp is in seconds.
Here's a screenshot of pcap POST request packet:
You can see that the precision used is ns instead of s.

__Specifications:__
- Client Version: 2.9.0
- InfluxDB Version: 2.6.1
- Platform: Alpine Linux v3.17

Start by tracing Client.new through create_write_api and the batching path used by WriteOptions, focusing on where the write precision becomes the POST request's query parameter. Reproduce the supplied seconds-precision batching configuration and verify that the request uses seconds rather than nanoseconds and that the data is accepted by InfluxDB.

Written by the indexing model from the issue text.