Conflicting guidance on lazy heavy imports

[rparkr]

Summary

STYLEGUIDE.md and lazy_heavy_imports.py give conflicting guidance on importing heavy third-party libraries. The two sources should be aligned to clarify the expected pattern to use when writing new code.

The style guide

STYLEGUIDE.md notes that large dependencies should be lazy-loaded by importing the library alias from the data_designer.lazy_heavy_imports module:

DataDesigner/STYLEGUIDE.md

Lines 145 to 147 in 7c2c15e

	# Use lazy loading with IDE support
	from typing import TYPE_CHECKING
	from data_designer.lazy_heavy_imports import pd, np

STYLEGUIDE.md also explains that any newly added heavy dependencies (e.g., your_lib) should use that same pattern:

DataDesigner/STYLEGUIDE.md

Line 176 in 7c2c15e

from data_designer.lazy_heavy_imports import your_lib

The module docstring

In lazy_heavy_imports.py, the docstring warns against the exact pattern that STYLEGUIDE.md recommended:

DataDesigner/packages/data-designer-config/src/data_designer/lazy_heavy_imports.py

Lines 19 to 21 in 7c2c15e

	Important:
	Avoid `from data_designer.lazy_heavy_imports import pd`.
	That import style resolves the attribute immediately and eagerly imports the heavy dependency.

Proposed resolution

Align both documentation sources to the same pattern to avoid confusion about which method is preferable, i.e., either importing the library alias from lazy_heavy_imports:

from data_designer.lazy_heavy_imports import pd, np 

or importing the lazy_heavy_imports module and accessing its attributes at runtime:

import data_designer.lazy_heavy_imports as lazy

# ...when needed in the code:
df = lazy.pd.DataFrame(...)