editoast: infra: order objects when downloading a railjson

[Castavo]

This PR is there to compensate the fact that railjson import loads objects (track sections, signals, ...) in a random order in the db.

Then, if you import and export a railjson, the order of the objects change randomly. This randomness affects the results of the circulation import benchmark (today if you import the same railjson twice, and run the circulation import benchmark on each, the results aren't the same). This PR tries to fix that.

I ran this code before and after.

INFRA_ID = 3  # France
url = f"http://localhost:8090/infra/{INFRA_ID}/railjson"

# first call
start = time()
response = requests.get(url)
first_call_time = time() - start

# run ten times
times = []
for _ in tqdm(range(100)):
    start = time()
    response = requests.get(url)
    times.append(time() - start)

Before:
first call: 2.50 seconds
mean: 2.11 seconds
stdev: 0.05 seconds

After:
first call: 2.65 seconds
mean: 2.34 seconds
stdev: 0.04 seconds

I'm not 100% certain that the results can be trusted, the performance of the endpoint seems very variable depending on the context.

So there is a bit more than a 10% of time increase with this PR, do you think it's acceptable ? If not, I can try adding indexes to the obj_id columns and run another benchmark.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 27.93%. Comparing base (689d6c3) to head (52d835e).

Additional details and impacted files

@@            Coverage Diff            @@
##                dev    #6728   +/-   ##
=========================================
  Coverage     27.92%   27.93%           
  Complexity     2172     2172           
=========================================
  Files          1041     1041           
  Lines        128661   128661           
  Branches       2567     2567           
=========================================
+ Hits          35928    35937    +9     
+ Misses        91250    91241    -9     
  Partials       1483     1483           

Flag	Coverage Δ
core	79.36% <ø> (ø)
editoast	75.38% <100.00%> (-0.01%)	⬇️
front	8.41% <ø> (ø)
gateway	2.50% <ø> (ø)
railjson_generator	87.26% <ø> (ø)
tests	83.07% <ø> (+1.12%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[flomonster]

LGTM