update Python version requirement and TensorFlow dependency in pyproj…

[bnb32]

Migrating to keras 3, tensorflow < 2.21, and python < 3.14

• Primarily motivated by flash attention being available for 2.16+
• Also, since Keras 3 splits from tensorflow and becomes a multi-backend library this could be a first step towards removing dep on tensorflow (tf.keras -> keras) and being able to use pytorch or JAX as a backend.
• Lots of back and forth on saving / loading models but ultimately we can keep that logic as long as we register custom layers with keras to be serialized.
• Explicit building of child layers also now required for correct deserialization.

Lots of auto-formatting changes but real net changes are fairly minimal, almost entirely in custom_layers.py, and consist of the following:

• register custom layers
• add get_config methods
• add build methods and child build calls.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 79.59184% with 60 lines in your changes missing coverage. Please review.
✅ Project coverage is 87.91%. Comparing base (54e9753) to head (525ceb9).

Files with missing lines	Patch %	Lines
phygnn/layers/custom_layers.py	65.24%	49 Missing ⚠️
phygnn/layers/handlers.py	54.54%	5 Missing ⚠️
phygnn/base.py	85.71%	3 Missing ⚠️
phygnn/phygnn.py	93.75%	2 Missing ⚠️
phygnn/model_interfaces/phygnn_model.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #72      +/-   ##
==========================================
- Coverage   88.36%   87.91%   -0.45%     
==========================================
  Files          24       24              
  Lines        3583     3559      -24     
==========================================
- Hits         3166     3129      -37     
- Misses        417      430      +13     

Flag	Coverage Δ
unittests	87.91% <79.59%> (-0.45%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull request overview

Updates project runtime requirements (Python/TensorFlow) and adjusts TensorFlow/Keras integration to remain compatible with newer TF/Keras behaviors (notably model serialization and layer/optimizer handling).

Changes:

• Bump supported Python baseline to 3.9+ and update TensorFlow pin to >=2.16.1,<2.17 (plus CI matrix updates).
• Update model save/load behavior to use a model.keras artifact and attempt custom-layer deserialization support.
• Keras-compatibility tweaks in core training/inference (optimizer deserialization, gradient tape usage, InputLayer handling) and removal of the FNO layer/tests.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
pyproject.toml	Raises Python requirement and updates TensorFlow dependency pins.
.github/workflows/pull_request_tests_tf_latest.yml	Updates PR test matrix to drop 3.8 and add 3.12 coverage.
.github/workflows/linter.yml	Removes super-linter job, leaving Ruff as the lint gate.
phygnn/model_interfaces/tf_model.py	Saves to model.keras and loads with discovered custom_objects.
phygnn/layers/custom_layers.py	Adjusts several custom layer constructors for Keras deserialization; removes FNO implementation.
phygnn/base.py	Updates prediction path to avoid calling Keras 3 InputLayer directly; formatting/logging changes.
phygnn/phygnn.py	Uses optimizers.deserialize for optimizer config; removes explicit tape variable watching.
tests/test_layers.py	Removes FNO tests; updates a nested hidden-layer config from Conv2D to Conv3D.

---

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

[Copilot]

Pull request overview

Copilot reviewed 9 out of 9 changed files in this pull request and generated 4 comments.

---

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

[Copilot]

Pull request overview

Copilot reviewed 9 out of 9 changed files in this pull request and generated 3 comments.

---

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

[Copilot]

Pull request overview

Copilot reviewed 13 out of 13 changed files in this pull request and generated 3 comments.

---

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

[bnb32]

@grantbuster There's not a ton to review here. Main "reviewable" is whether any breaking changes from keras 3 migration are worth caring about. I'll interpret emoji reactions and / or silence as approval :)