Humioapi

Changed

- Changed all syncronous requests through `httpx` to use urllib3 as networking backend rather than `httpcore` as a
temporary solution, since the humio-search-all and graphql endpoints started giving random HTTP 502s since Humio 1.18.
Will probably revert in the future when I can figure out whats up.
- Changed the `humioapi.loadenv()` helper to accept iterables of envs and prefixes as well as the old strings.

Changed

- Renamed `async_streaming_tasks` to `async_streaming_search` and reworked the API to make it easier to setup multiple
queries with different properties by passing queries as a list of dicts.
- Bumped dependency versions

Fixed

Some README typos

Fixed

- Added color style to TRACE logging level in console renderer logging helper, since httpx defines that level (5).

Added

- Optional timeout parameters to search APIs. By default 30 seconds. Can be an integer og a httpx.Timeout.

Changed

- Refactored logging helpers with a new function `humioapi.initialize_logging` providing sane structlog JSON (default) or Human readable logs depending on the chosen format.
- QueryJob now takes a client instance instead of a token and base_url
- Bumped dependency versions

Fixed

- Use context manager with all `httpx` requests so connections are closed properly.

Removed

- `humioapi.setup_excellent_logging` removed in favor of `humioapi.initialize_logging`

Fixed

- Renamed `end` to `start` in a few forgotten places (url util functions)
- Strip trailing slash in base urls instead of stupidy failing

This release renames the module and repository to humioapi to better
communicate what this python module actually is.

Added

- The `humioapi.async_streaming_tasks` function which constructs asyncronous HTTP streaming search tasks in the form of asyncronous generators.
- The `humioapi.consume_async` helper function which can be used to consume the asyncronous streaming tasks from a non-asyncronous context as if they were a ordinary python generator.

Changed

- Removed `requests` in favor of `httpx`
- All `start` and `stop` parameters now accept multiple timestamps, which will be parsed with `humioapi.parse_ts`. In other words they can be `millisecond-epochs` (as Humio likes it), relative Splunk-style `snaptime tokens`, `iso8601 time strings` (or any other time format supported by `pendulum.parse`).
- Renamed `end` to `stop` in all API functions, since `start/stop` sound better to me than `start/end`, plus this has the bonus effect that the fields are sorted near each other in logging output :)
- Made `pandas` optional (its only used in the POC/playground timeseries after all)
- All time parsing that previously resulted in a `pandas.Timestamp` now use `pendulum.Datetime` instead.

Fixed

- Better defaults in `humioapi.QueryJob`

Removed

- Renamed `humiocore` to `humioapi`.