manager.generate_report fails in CLI mode

[deniederhut]

Describe the bug
When running curifactory in CLI mode, generating a report inside the run causes a failure

To Reproduce
Here is the experiment definition, inside an experiments folder:

from dataclasses import dataclass

from curifactory import ArtifactManager, Record, stage
from curifactory.caching import Lazy, PickleCacher


@stage(inputs=None, outputs=[Lazy("initial_value")], cachers=[PickleCacher])
def get_initial_value(record):
    my_value = 5
    return my_value * record.params.number

@stage(inputs=["initial_value"], outputs=["final_value"], cachers=[PickleCacher])
def multiply_again(record, initial_value):
    return initial_value * record.params.number

def run(param_sets, manager):
    for param_set in param_sets:
        record = Record(manager, param_set)
        result = multiply_again(get_initial_value(record))
    manager.generate_report()

with params in a params directory that look like this

from dataclasses import dataclass

from curifactory import ExperimentParameters


@dataclass
class MyParams(ExperimentParameters):
    number: float = 1.0

def get_params():
    return [MyParams(number=number) for number in (0, 1, 2, 2)]

Running this with experiment run -p basic_params fails with

Traceback (most recent call last):
  File "/Users/dillon/miniconda3/envs/curifactory-dev/bin/experiment", line 8, in <module>
    sys.exit(main())
  File "/Users/dillon/miniconda3/envs/curifactory-dev/lib/python3.9/site-packages/curifactory/experiment.py", line 1380, in main
    run_experiment(
  File "/Users/dillon/miniconda3/envs/curifactory-dev/lib/python3.9/site-packages/curifactory/experiment.py", line 756, in run_experiment
    mngr.generate_report()
  File "/Users/dillon/miniconda3/envs/curifactory-dev/lib/python3.9/site-packages/curifactory/manager.py", line 574, in generate_report
    reporting.run_report(self, self.reports_path, "_latest", self.report_css_path)
  File "/Users/dillon/miniconda3/envs/curifactory-dev/lib/python3.9/site-packages/curifactory/reporting.py", line 655, in run_report
    html_lines.extend(render_report_stage_map(manager, graphs_path))
  File "/Users/dillon/miniconda3/envs/curifactory-dev/lib/python3.9/site-packages/curifactory/reporting.py", line 540, in render_report_stage_map
    graph = map_full_svg(manager)
  File "/Users/dillon/miniconda3/envs/curifactory-dev/lib/python3.9/site-packages/curifactory/reporting.py", line 933, in map_full_svg
    _add_record_subgraph(dot, index, record, manager, detailed)
  File "/Users/dillon/miniconda3/envs/curifactory-dev/lib/python3.9/site-packages/curifactory/reporting.py", line 696, in _add_record_subgraph
    if artifact.init_record == record:
AttributeError: 'MapArtifactRepresentation' object has no attribute 'init_record'

Expected behavior
If the report has already been generated, I think I would expect to not generate a second identical one?

[WarmCyan]

Huh, I think this is actually because it's trying to generate the report while in map mode, I'll need to add a check for that in the various generation functions, although it's a good point that I should also add a flag to avoid a duplicate generation if run in non-dag mode or once mapping has completed.