Merge pull request #33 from brainglobe/dev-docs-draft

Clarify repo purposes

README.md

@@ -23,6 +23,17 @@ At present, the package offers the following workflows:
 
 - [cellfinder](#cellfinder): Whole-brain detection, registration, and analysis.
 
+These workflows should be representative of the most common use-cases and are meant to be easy to reuse. They also serve as an example of how to combine several BrainGlobe tools to achieve a goal, such as whole brain cell detection and atlas registration.
+These workflows typically combine several BrainGlobe tools (possibly together with other tools) to achieve a goal,
+such as whole brain cell detection and atlas registration.
+
+## Secondary purposes of brainglobe-workflows, for developers
+
+We also use these workflows to support code development. We do this by regularly benchmarking the time they take to complete to ensure performance is stable as the code changes.
+* Developers can install these benchmarks locally via `pip install [dev]`. By executing `asv run`,  the benchmarks will run with default parameters on a small dataset that is downloaded from [GIN](https://gin.g-node.org/G-Node/info/wiki). See [the asv docs](https://asv.readthedocs.io/en/v0.6.1/using.html#running-benchmarks) for further details on how to run benchmarks.
+* Developers can also run these benchmarks on  data available locally, by specifying the relevant paths in an input configuration file (JSON file).
+* We additionally run and benchmark the workflows locally on a internal desktop machine with large example datasets. These benchmarks are run periodically and the results are made publicly available.
+
 ## Installation
 
 If you want to install BrainGlobe workflows as a standalone tool, you can run `pip install` in your desired environment: