[BUG]: problem with python and c++ data type conversion

Required prerequisites

• Make sure you've read the documentation. Your issue may be addressed there.
• Search the issue tracker and Discussions to verify that this hasn't already been reported. +1 or comment there if it has.
• Consider asking first in the Gitter chat room or in a Discussion.

What version (or hash if on master) of pybind11 are you using?

2.10.4

Problem description

I try to create an ndarray(whose size is 5, and dtype is float) in c++ interface as the return value, which will be passed to a python function as output.
I tried this ↓

and this ↓

But all the methods turns out that the python function output data only retains one digit after the decimal point，
which does not match the precision of float.
For example:
the return_data[0]=0.0025783
but the arr[0]=0.0

Where did I make a mistake?
Anyone gets any ideas??

Reproducible example code

No response

Is this a regression? Put the last known working version here if it is.

Not a regression

No repository file, test, entry point, or reproducible example is provided; start by reducing the reported C++/Python ndarray conversion to a minimal example using pybind11 2.10.4. Compare the returned values and dtypes, then establish whether the behavior is a pybind11 defect or an issue in the omitted example before proposing a fix or documentation update.