Add M/M/c queue example as canonical DES model

[EwoutH]

Summary

Adds a classic M/M/c queuing model as the first pure discrete event simulation example, demonstrating Mesa's new event scheduling API without using step().

Motive

Mesa 3.5 introduced public event scheduling and time advancement (mesa/mesa#3266), but we lack an example showing pure DES usage. The M/M/c queue is the "hello world" of discrete event simulation: simple, well-understood, and with analytical solutions for validation. It fills the gap between our ABM examples and the new DES capabilities.

Implementation

• agents.py: Customer (created on arrival, removed after service) and Server (pulls from queue when idle — server-centric design).
• model.py: MMcQueue model using schedule_recurring with a callable interval for Poisson arrivals and schedule_event(after=...) for service completions. Disables the default step schedule via _default_schedule.stop().
• analytical_mmc.py: Erlang C steady-state solutions for validation.

Usage Examples

from model import MMcQueue, analytical_mmc

model = MMcQueue(arrival_rate=2.0, service_rate=1.0, n_servers=3, rng=42)
model.run_until(10_000)

print(model.server_utilization)  # ≈ 0.667
print(model.avg_wait_time)       # ≈ 0.375

# Compare against analytical steady-state values
analytical = analytical_mmc(2.0, 1.0, 3)

Running python model.py prints a comparison table:

M/M/3 Queue (λ=2.0, μ=1.0, T=10000.0)
Customers served: 19992

Metric                     Simulated   Analytical
---------------------------------------------------
Server utilization              0.6672       0.6667
Avg wait time                   0.3716       0.3750
Avg system time                 1.3716       1.3750

Additional Notes

• The model disables the default step schedule with _default_schedule.stop(). This works but may warrant a more explicit opt-out mechanism for pure DES models (e.g. a Model parameter).
• No visualization app is included yet, an implementation would be appreciated!

[quaquel]

consider using a scenario class here.

[quaquel]

this is a question beyond this example, but should this not become some seperate mesa base class we can use?

[EwoutH]

Could very well be!

I mainly created this example to see if we supported everything to do it. The answer is yes, and for future potential Mesa improvements we could check if it simplifies this example or not.

[quaquel]

this also reveals something we have to fix more elegantly in mesa 4.

[EwoutH]

Agreed.

[quaquel]

what about adding the new data collection to this example. Your use of properties should make this quite easy.

[EwoutH]

Sounds cool, but let's make that a separate PR.