Validate CSV output schemas

Also moved code dir to src.
There are reasons to do standard things in standard ways. While it is
possible to get the `code/` directory to work, and recognize it as a
package path, this requires wrangling the pyproject.toml file.
Additionally, any import from the `code.something` path automatically
shadows the python stdlib `code` module. While it may not be necessary,
it still is good to not shadow standard library modules.

README.md

@@ -9,7 +9,7 @@ Any action can easily be started using [`just`](https://github.com/casey/just) w
 ## Dataset structure
 
 - All inputs (i.e. building blocks from other sources) are located in `input/`.
-- All custom code is located in `code/`.
+- All custom code is located in `src/`.
 - All final output data is located in `output/`
 
 ## Output data structure
@@ -51,7 +51,7 @@ Contained in `packages.csv`, 4 columns:
 
 Represents information about the unique system installations represented in the raw dataset.
 
-Contained in `packages.csv`, 2 columns:
+Contained in `unique_installs.csv`, 2 columns:
 
 - `date`: the date a specific file is relevant for
 - `unique`: the amount of unique installations counted on the observation date