Closes #5441: Remove deprecated modules

[ajpotts]

Summary

This PR completes a structural refactor of the Arkouda Python package
by:

• Removing deprecated top-level legacy modules.
• Redirecting imports to their canonical locations under:
  • arkouda.pandas
  • arkouda.numpy
  • arkouda.core
  • arkouda.scipy
• Updating documentation, tests, and internal imports to reflect the
  new structure.

The goal is to eliminate transitional shims and finalize the namespace
reorganization.

---

Motivation

Historically, Arkouda exposed many modules at the top level (e.g.,
arkouda.dataframe, arkouda.groupbyclass, arkouda.util, etc.).
Over time, these were reorganized into clearer logical namespaces:

• Pandas-like APIs → arkouda.pandas.*
• NumPy-like APIs → arkouda.numpy.*
• Core client/server utilities → arkouda.core.*
• Sparse/scipy-related → arkouda.scipy.*

Deprecated compatibility modules remained as thin wrappers issuing
DeprecationWarnings.
This PR removes those wrappers and updates all references to their
canonical modules.

---

What Changed

1. Removed Deprecated Top-Level Modules

Deleted legacy modules such as:

• arkouda.dataframe
• arkouda.groupbyclass
• arkouda.index
• arkouda.categorical
• arkouda.series
• arkouda.util
• arkouda.sorting
• arkouda.random
• arkouda.segarray
• arkouda.strings
• arkouda.timeclass
• arkouda.io
• arkouda.io_util
• arkouda.row
• arkouda.accessor
• arkouda.numeric
• arkouda.dtypes
• arkouda.match, arkouda.matcher
• arkouda.message
• arkouda.security
• arkouda.logger
• arkouda.sparrayclass
• arkouda.sparsematrix
• arkouda.client
• arkouda.pdarrayclass
• arkouda.pdarraycreation
• arkouda.pdarraymanipulation
• arkouda.pdarraysetops
• arkouda.alignment
• arkouda.infoclass
• arkouda.join

These modules previously re-exported symbols from their new locations
and emitted DeprecationWarnings.

---

2. Canonicalized Imports

All internal imports were updated to use the correct module paths, for
example:

Old Import New Import

---

arkouda.dataframe arkouda.pandas.dataframe
arkouda.groupbyclass arkouda.pandas.groupbyclass
arkouda.index arkouda.pandas.index
arkouda.categorical arkouda.pandas.categorical
arkouda.util arkouda.numpy.util
arkouda.sorting arkouda.numpy.sorting
arkouda.pdarraycreation arkouda.numpy.pdarraycreation
arkouda.infoclass arkouda.core.infoclass
arkouda.logger arkouda.core.logger

This affects: - Core library code - Pandas extension API - Tests -
Documentation - Training materials

---

3. Documentation Updates

Updated: - Sphinx autodoc references - Markdown documentation - Usage
guides - File I/O documentation - Training materials

All references now consistently point to:

• arkouda.pandas.DataFrame
• arkouda.pandas.Series
• arkouda.pandas.Index
• arkouda.pandas.Categorical
• arkouda.numpy.* utilities

---

4. Tests Updated

All tests were updated to import from canonical module locations.
Assertions that referenced old module paths were updated accordingly.

---

Behavioral Impact

• ❗ Code importing deprecated modules (e.g., arkouda.dataframe)
  will now fail instead of emitting a warning.
• ✅ No runtime behavior of core functionality is changed.
• ✅ No algorithmic or performance changes.
• ✅ Public API remains available via canonical namespaces.

This is a namespace cleanup only.

---

Benefits

• Eliminates confusing duplicate module paths.
• Clarifies architectural boundaries:
  • Pandas API layer
  • NumPy API layer
  • Core infrastructure
• Reduces maintenance burden.
• Prepares the project for future API stabilization work.

---

Migration Guide

Replace old imports:

# OLD
from arkouda.dataframe import DataFrame
from arkouda.groupbyclass import GroupBy
from arkouda.util import register

# NEW
from arkouda.pandas.dataframe import DataFrame
from arkouda.pandas.groupbyclass import GroupBy
from arkouda.numpy.util import register

If using sparse functionality:

from arkouda.scipy.sparsematrix import create_sparse_matrix

---

Conclusion

This PR finalizes the Arkouda namespace refactor by removing deprecated
compatibility modules and standardizing all imports across the codebase.
It simplifies the package structure and clarifies the architectural
layering between pandas-style APIs, numpy-style APIs, and core
infrastructure.

Closes #5441: Remove deprecated modules

[jaketrookman]

Looks good