Add framework for graph streaming, ray implementation

[amitschang]

To overcome some drawbacks of ray data pipelines for the kinds of workloads we want:

• No support for streaming split. In the previous project, we wanted to have large batches for one process which fed to one that desired small batches. This could be faked, but at the expense that a write-out would wait for the whole batch, decreasing resilience
• No support for multiple inputs, or branching (Ray data only supports simple-path graphs). This could be simulated at the task level, but that leads to higher complexity and less portability in task code.
• Custom generators. These can be done in ray data but come with some drawbacks (also below)
• Indefinite generators. Ray data is designed more for data table ETL and less for streaming, so requires knowledge of total number of rows.

A few things that are out of scope here, but can be more easily integrated:

• Task result caching for resumption
• backpressure
• Implementation in different remote frameworks (example: parsl). The stream class handles most logic, the implementation just needs to manage resources, start tasks, and process results (e.g. see local serial implementation)

Notes:

• This ought to be backward compatible with the existing executor (it promotes list-of-tasks to graph)
• There is no join concept, multiple edges to a single task proceed in a package-sorter type of fashion. This supports tasks that are separate configurations of the same task, or tasks that produce compatible data products (e.g. different protein folding methods), and different outputs.
• Docs will be added in a separate PR - post-review

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Comparison is base (d691281) 100.00% compared to head (a3b4529) 100.00%. Report is 3 commits behind head on main.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #43 +/- ## ========================================== Coverage 100.00% 100.00% ========================================== Files 9 11 +2 Lines 269 551 +282 ========================================== + Hits 269 551 +282 ``` | [Flag](https://app.codecov.io/gh/ssec-jhu/dplutils/pull/43/flags?src=pr&el=flags&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=ssec-jhu) | Coverage Δ | | |---|---|---| | [unittests](https://app.codecov.io/gh/ssec-jhu/dplutils/pull/43/flags?src=pr&el=flag&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=ssec-jhu) | `100.00% <100.00%> (ø)` | | Flags with carried forward coverage won't be shown. [Click here](https://docs.codecov.io/docs/carryforward-flags?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=ssec-jhu#carryforward-flags-in-the-pull-request-comment) to find out more.

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[jamienoss]

On the case, but looks like there's a fair bit to review.

[amitschang]

On the case, but looks like there's a fair bit to review.

cool, take your time! Sorry bout the size, but only way it made sense to me was to bring it all together

[amitschang]

thanks y'all!