fix: Use correct model for low stats demonstration

[matthewfeickert]

Resolves #63

• Use the correct model (model_low) for demonstration of the difference between the asymptotics based and toy based calculators.
  • Amends PR feat: Add tutorial skeleton from LHC Reinterpretation Forum Workshop 2021 #1
• Use 5000 pseudo-experiments to demonstrate the difference more effectively.
• Use underscores in numeric literals (5_000) for better readability.
  • c.f. https://peps.python.org/pep-0515/

* Use the correct model (`model_low`) for demonstration of the
  difference between the asymptotics based and toy based calculators.
   - Amends PR https://github.com/pyhf/pyhf-tutorial/pull/1
* Use 5000 pseudo-experiments to demonstrate the difference more
  effectively.
* Use underscores in numeric literals (5_000) for better readability.
   - c.f. https://peps.python.org/pep-0515/

[matthewfeickert]

Just above this is

model_low = pyhf.simplemodels.uncorrelated_background(
    signal=[0.5, 1.0], bkg=[5.0, 6.0], bkg_uncertainty=[0.5, 1.2]
)

but we were never using this model_low for some reason and so were trying to fit a model that terribly modeled the observations to begin with. This PR just used the model we meant to.