Break up main into separate files and added CPU timeline

.gitmodules

@@ -1,3 +0,0 @@
-[submodule "ILLIXR"]
-	path = ILLIXR
-	url = [email protected]:ILLIXR/ILLIXR.git

README.md

@@ -1,5 +1,19 @@
-- The exact code I used is in [`ILLIXR`](./ILLIXR) (submodule)/.
-- An exact copy of the metrics I recovered are available in [`metrics`](./metrics), so this analysis is reproducible.
-- Read [`trial_conditions.yaml`](./trial_conditions.yaml) for trial conditions/environment.
-- To run the analysis yourself, see [`./analysis/README.md`](./analysis/README.md).
-- Read [`output/README.md`](./output/) for interesting things in the output.
+# Adding Data
+1. Follow our [Getting Started](https://illixr.github.io/ILLIXR/getting_started/) guide to run ILLIXR.
+    * If you want meaningful results, set `profile: opt` in the `config/$YOUR_CONFIG.yaml` where `$YOUR_CONFIG.yaml` is the config you want to run for your experiment.
+2. Running `cd ILLIXR ; ./runner.sh config/$YOUR_CONFIG.yaml` should create a directory called `ILLIXR/metrics/`. Move this directory to `results/metrics/$NAME` where `results` is this repo and `$NAME` is an arbitrary unused directory name.
+3. Add metadata to `metrics/$NAME/trial_conditions.yaml`. See `analysis/util.py:TrialConditions` for the schema of this YAML. You can add extra fields if you have extra experimental conditions to track; I will always encourage people to embed machine-readable labels their data :)
+
+# Running the Script
+1. If this is your first time invoking thi script, install the dependencies by `cd results/analysis ; poetry install`.
+2. Run `cd results/analysis ; poetry run python3 main.py`. This will run all of the analyses on all of the data.
+3. Examine the results in `output/`. Note that `output/$NAME` corresponding to `metrics/$NAME`. `output/account_summaries.md` is a very good place to start.
+
+# Adding an Analysis or Plot
+- Per-trial and inter-trial analyses can be found in the `analysis/per-trial-analysis.py` and `analysis/inter-trial-analysis.py` files, respectively.
+- Per-trial analyses takes a single `PerTrialData`, while inter-trial analyses take a list of `PerTrialData`. See `analysis/util.py:PerTrialData` for what each attribute means.
+- The `analysis` functions in each file run all inter-trial or per-trial analysis within.
+- We've found it helpful to comment out the slow ones.
+- Running the analyses is a slow way of revealing errors; To speed up development, you can quickly type check your analyses using Mypy,
+  - Run `cd results/analysis ; poetry install --dev` and `poetry run mypy --strict --ignore-missing-imports main.py`.
+  - Use `:` and `->` to signify the parameter-types and return-type of new functions (e.g. `def foo(x: int, y: float) -> str`).