Add API methods for profiling pipeline execution

[tdruez]

The goal is to locate the bottlenecks for future optimizations.

[tdruez]

A profile() method is now available on the Run model for profiling pipeline execution:

from scanpipe.models import Run

run = Run.objects.get(project__uuid="6ab92dfe-8623-4753-8fb6-98c9c9ae0a5b")
run.profile()

{'analyze_scanned_files': 1,
 'collect_and_create_codebase_resources': 73,
 'collect_and_create_system_packages': 525,
 'collect_images_information': 1,
 'end': 1,
 'extract_images': 5,
 'extract_layers': 4,
 'find_images_linux_distro': 1,
 'scan_for_application_packages': 1,
 'scan_for_files': 21,
 'start': 1,
 'tag_not_analyzed_codebase_resources': 1,
 'tag_uninteresting_codebase_resources': 4}

Formatted results can also be printed to stdout:

run.profile(print_results=True)

[tdruez]

Once a problematic step is located, https://github.com/joerick/pyinstrument is a nice tool to go deeper in the profiling:

pip install pyinstrument

from pyinstrument import Profiler

with Profiler() as profiler:
    code_to_profile

# Stdout
print(profiler.output_text(unicode=True, color=True))
# HTML
profiler.open_in_browser()