During black box testing of FR5 (View Workout) a bug in the class-level
validation on the UserSerializer class was discovered. When accepting a
coach offer, user validation would fail, complaining about the password
being None.

The changes made to the initial implementation is know mostly reverted.
Class-level validation is removed in favour of field-level validation
for the password field. The implementation of the password validation,
however, remains the same.

Additionally, default values for related required fields are know
declared using the `extra_kwargs` attribute.

Unit tests for the UserSerializer class is also updated, to reflect the
changes.

Update CI setup for backend

See merge request !9

- Include cobertura coverage report
- Setup cache and share it between branches
- Remove dependencies introduced to provide pg_config

Backend/testing/user serializers

See merge request !8

- Do custom validation on object-level instead of field-level
- Use Django password validators to validate password
- Explicitly validate request POST data when creating new users

Feature/highscores rank by units

See merge request !7

When viewing high scores for a specific exercise, it is now also
possible to rank the high score list by total units performed.

- The high score list is now rendered as a table
- For exercises a new column showing units performed is visble
- For workouts only the score column is visible
- The high score list is sorted by both score and units in the backend

Honour active list view when filtering on exercise

Closes #11

See merge request !6

When filtering workouts on exercise, the filter should be applied on the
active workout list view. Prior to this fix, the filter was applied to
all workouts regardless of what workout filter tab was selected by the
user.

Closes #11.

The "good old copy and paste" messed up the query on element ID in
workouts.js. The select element used for filtering on exercise now uses
the same ID name for both workouts and highscores.

Also check the response status before accessing response data.

Closes #10.

The label text and the id attribute of select element is also updated to
better match the context.

Closes #9.

Make the pg_config available on test image

See merge request !5

The psycopg2 requirement got included in the requirements.txt file when
django-heroku was added in commit 1667ceda, which in turn broke the CI
pipeline.

- Use django-heroku to configure backend for heroku
- Use dj-database-url to configure database with DATABASE_URL
- Add Procfile for backend
- Add requirements.txt to repo root for heroku/python build
- Move runtime.txt to repo root
- Add Procfile for frontend
- Add package.json to repo root for heroku/nodejs build

Using pipenv to install dependencies from requirements.txt resulted in
the following error (pipenv install -r requirements.txt):

    Unexpected syntax in …/requirements.txt.
    Are you sure this is a requirements.txt style file?
    'utf-8' codec can't decode byte 0xff in position 0: invalid start byte

Turns out the encoding of the file got messed up somehow!?
The invalid start byte is caused by the <feff> character.
Running file requirements.txt reveals:

    requirements.txt: Little-endian UTF-16 Unicode text, with CRLF line terminators

Fixed by running:

    iconv -f utf-16 -t utf-8 requirements.txt -o requirements.txt.new
    dos2unix requirements.txt.new
    mv requirements.txt.new requirements.txt

Add profile page 😎

See merge request !4

Merge feature/highscore-list

Closes #3 and #1

See merge request !3

It is now possible to filter the high score list on exercise types using
a query parameter in the api call.

Created a html page for highscores.
It renders a container with a title and a description, and a list of highscores.
It also has a template for rendering highscore components to the list.

Added new javascript file, highscores.js.
It adds an event listener to the window, which triggers when DOM is fully loaded.
The event listener fetches all highscores and renders all highscores stored.
If response is not successfull, a feedback is displayed to the user.

The function fetchHighscores sends a request to the api enpoint highscores.
If response is successfull, a component from the highscore template is rendered with highscore information.

Added a line in navbar.js, rendering a button on the navigation bar redirecting the user to highscore view.

Added a new statement to updateNavBar function in scripts.js.
The statement checks if the browser window has a pathname for highscores.html, and activates the link redirecting to the page if it exists.
If user is authenticated, the css class hide is removed from the DOM element.

Added some new styles to style.css used to design highscore.html view.

Add filter on exercise to workouts

See merge request !2

Backend/testing/permissions

See merge request !1

Two new tests added to secure full test coverage.

Writing tests to get full coverage of permission.py in workouts folder in backend.