Configuration option to drop oldest or newest data in the buffer when buffer is full

### Use Case

Similar to https://github.com/influxdata/telegraf/issues/15208

(accommodating temporary high-write bursts to and influxdb instance)

Sometimes the newest data is more important to keep than the oldest data (if your computer failed, would you care more about what happened right before the failure or 5 minutes before the failure?)

Sometimes the oldest data is more important to keep than the newest data (ex: which events happened first in this very large burst of events?)

### Expected behavior

I expected to see a configuration option like "buffer_drop_strategy" which would default to "drop_oldest".

### Actual behavior

No configuration option to specify which metrics (newest or oldest) are dropped from the buffer.

### Additional info

_No response_

No files, tests, or entry points are named. Start by locating the buffer implementation and its configuration handling, then determine how a buffer-full condition selects metrics to discard; done means a configurable drop strategy supports both oldest and newest data, with the documented default behavior covered by tests.

Written by the indexing model from the issue text.