fix(chen2021Fitness): don't hand out invalid JSON when there are not enough data to compute a fit

[fengelniederhammer]

Some values were NaN or Infinity, which are no valid JSON tokens. Instead throw an error.

resolves #41

This PR tries to be minimally invasive, as this repo is mostly in maintenance mode. I only added what is necessary to fix the issue. As this repo has no test coverage, I don't want to risk more than necessary.

How I tested this locally:

• I had to update the dependencies in the model to run the Python code. My requirements.txt looks as follows:

click==7.1.2
Flask==1.1.2
gunicorn==20.0.4
itsdangerous==1.1.0
Jinja2==2.11.3
MarkupSafe==1.1.1
numpy==1.22.3
pandas==1.4
patsy==0.5.6
python-dateutil==2.8.1
pytz==2021.1
scipy==1.8.0
six==1.15.0
statsmodels==0.14.4
Werkzeug==1.0.1

• Change MODEL_ENDPOINT = "http://localhost:7070"
  

  cov-spectrum-server/src/main/kotlin/ch/ethz/covspectrum/controller/Chen2021FitnessController.kt

  Line 15 in f981d5a

  const val MODEL_ENDPOINT = "https://cov-spectrum.org/api/model-chen2021fitness"

• Run the Kotlin code with env variables COV_SPECTRUM_DB_USERNAME=dummy;COV_SPECTRUM_DB_PASSWORD=dummy;COV_SPECTRUM_JWT_TOKEN_LIFETIME_SECONDS=10
• Comment out the flyway migration
  

  cov-spectrum-server/src/main/kotlin/ch/ethz/covspectrum/CovSpectrumServerApplication.kt

  Lines 15 to 16 in f981d5a

  	val flyway = Flyway.configure().dataSource(getDbJdbcUrl(), getDbUser(), getDbPassword()).load()
  	flyway.migrate()

• Run the code, check which class complains about missing env variables, comment out the whole class, run the code...

[JonasKellerer]

LGTM