ENH: switch to copy=False by default when DataFrame is constructed from dict of Series/Index for CoW

[phofl]

Feature Type

• Adding new functionality to pandas
• Changing existing functionality in pandas
• Removing existing functionality in pandas

Problem Description

This would make stuff faster and CoW avoids that we mutate more than one object at once.

We should still copy arrays when they are part of the dictionary.

cc @jorisvandenbossche thoughts?

Feature Description

See above

Alternative Solutions

Continue copying

Additional Context

No response

[jorisvandenbossche]

For me that would be fine. As you mention, it shouldn't impact copy/view behaviour since with CoW we still track those columns were created from the Series/Index objects.
I think the main impact is that this will no longer result in a consolidated DataFrame (which I think is the main reason that this has a default of copy=True at the moment)

[phofl]

You could still set copy=True manually if this is important I guess? But yeah this is a fair point