docs: align API stability guarantees with 1.0.0 release (#597)

[sshekhar-04]

Overview
This Pull Request resolves the contradictions pointed out in issue #597 regarding the API stability status of the TensorFlow Java project.

While the 1.0.0 release notes officially committed the project to the TensorFlow API stability guarantees, several key documentation files (install.md and package-info.java) still contained "Experimental" warnings and disclaimers. These legacy warnings often discourage production adoption and conflict with the project's current maturity level.

Changes Made

1. tensorflow-core-api/.../package-info.java:

• Removed the WARNING label describing the API as experimental.
• Added an API Stability section that confirms adherence to the 1.0.0 stability guarantees.
• Included a nuanced disclaimer noting that stability is subject to the evolution of the upstream TensorFlow C API, addressing the concerns raised by @Craigacp.

2. docs/install.md:

• Replaced the high-visibility Caution box with a professional Note regarding the stability policy.
• Ensured the wording matches the updated package-info.java for consistency across the repository.

Technical Nuance
As discussed in the issue thread, this PR does not offer a "blanket" guarantee that ignores the realities of the underlying C++ core. Instead, it shifts the narrative from "Experimental/Unstable" to "Stable with Upstream Dependencies," which accurately reflects the architecture of the Java bindings.

CC: @Craigacp @karllessard @mihaimaruseac Note: I've updated the wording to find a middle ground between the 1.0.0 release guarantees and the upstream dependency concerns discussed in #597.

Related Issues
Fixes #597

[mihaimaruseac]

This seems in alignment with #597 but project maintainers should be the ones approving.