feat(python/sedonadb): Expose memory pool and runtime configuration in Python bindings

[Kontinuation]

Summary

• Add SedonaContextBuilder in the sedona Rust crate to centralize runtime environment construction (memory pool, disk manager) so it can be reused across CLI, Python, ADBC, and future bindings.
• Add runtime configuration properties (memory_limit, temp_dir, memory_pool_type, unspillable_reserve_ratio) to the Python Options class, allowing Python users to configure memory pools (greedy/fair) and disk spill settings for out-of-core workloads.
• Implement deferred context initialization in Python: the internal Rust context is created lazily on first query, so users can set runtime options via sd.options before execution begins.
• Simplify InternalContext::new() to accept a single HashMap<String, String>, delegating to SedonaContextBuilder::from_options.
• Add shared size_parser module so memory_limit accepts human-readable strings (e.g., "4gb", "512m") across CLI and Python.
• Initialize env_logger in the Python module for debugging support.
• Minor cleanup: switch log dependency in sedona-geo-generic-alg to workspace version.

Motivation

The memory pool and runtime environment configuration added in #599 was only accessible via sedona-cli. This PR exposes the same configuration through the Python bindings so that Python users can control memory limits and spill behavior when running out-of-core spatial joins.

The SedonaContextBuilder centralizes pool/runtime construction so it can be shared across CLI, Python, ADBC, and future R bindings without duplicating logic.

Example usage

import sedona.db

sd = sedona.db.connect()
sd.options.memory_limit = "4gb"
sd.options.temp_dir = "/tmp/sedona-spill"
sd.options.memory_pool_type = "fair"
sd.options.unspillable_reserve_ratio = 0.2

# Options are frozen once the first query triggers context creation
df = sd.sql("SELECT ST_Point(1.0, 2.0)")

[Copilot]

Pull request overview

This PR exposes memory pool and runtime environment configuration to Python users through the SedonaContext and connect() functions. These configuration options were previously only available via the sedona-cli tool (added in PR #599). The changes allow Python users to configure memory limits, disk spill behavior, and choose between greedy and fair memory pool types for out-of-core spatial joins.

Changes:

• Added memory pool configuration parameters (memory_limit, temp_dir, memory_pool_type, unspillable_reserve_ratio) to SedonaContext.__init__() and connect() function
• Initialized env_logger in the Python module for debugging support
• Updated sedona-geo-generic-alg to use workspace version of log dependency

Reviewed changes

Copilot reviewed 5 out of 6 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
python/sedonadb/src/context.rs	Adds runtime configuration parameters to InternalContext::new() and builds custom RuntimeEnv with memory pool and disk manager
python/sedonadb/python/sedonadb/context.py	Adds configuration parameters to SedonaContext.__init__() and connect() with type hints and documentation
python/sedonadb/src/lib.rs	Initializes env_logger for debugging support
python/sedonadb/Cargo.toml	Adds env_logger dependency
rust/sedona-geo-generic-alg/Cargo.toml	Updates log dependency to use workspace version
Cargo.lock	Reflects dependency changes

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[paleolimbot]

Just a few high-level notes that I think affect how this is structured:

• We'll need to do this in the ADBC driver, the CLI, in R, and in Python, so if this can be done mostly based on HashMap<String, String> and implemented in the sedona crate we can reuse the logic in all four places. We may want a SedonaContextBuilder given the growing complexity of assembling the pieces.
• In Python, the connect() function that is actually used lives in apache-sedona and creating a SessionContext() manually is not something anybody should really do. Probably implementing this via sd.options.xxx and deferring the creation of the local InternalContext so that these options can be set before the Rust context is created would work best.

[Copilot]

Pull request overview

Copilot reviewed 11 out of 12 changed files in this pull request and generated 4 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[paleolimbot]

Thank you for these updates! I'll follow up with the ADBC and R interfaces for setting the memory limit.

I re-ran CI (I think the previous failure was transient)