build: Add release_version.h to control the TF wheel filename and maintain reproducible wheel content and filename results.

It enables ability to pass defines values to `//tensorflow/core/public:release_version`.

Instead of having TF version in three places (`setup.py`, `version.h` and `tensorflow.bzl`), it should be located in one place only. Declaration of TF version in a new `tf_version.bzl` file will allow passing this value to build rules wrapping `setup.py` and `release_version.h`.

Dependency on `//tensorflow/core/public:release_version` should be added if `TF_MAJOR_VERSION, TF_MINOR_VERSION, TF_PATCH_VERSION, TF_VERSION_SUFFIX` values are used in the code.
Dependency on `//tensorflow/core/public:version` should be added if graphDef compatibility versions or checkpoint compatibility versions are used in the code.

The next step would be to change cc_library release_version in the following way:

```
cc_library(
    name = "release_version",
    hdrs = ["release_version.h"],
    defines = [
        "TF_MAJOR_VERSION={}".format(MAJOR_VERSION),
        "TF_MINOR_VERSION={}".format(MINOR_VERSION),
        "TF_PATCH_VERSION={}".format(PATCH_VERSION),
        "TF_VERSION_SUFFIX={}".format(TF_SEMANTIC_VERSION_SUFFIX),
    ],
    visibility = ["//visibility:public"],
)
```

The version chunks will be created from the value in `tf_version.bzl`. The version suffix will be created by a new repository rule, and it will be controlled by environment variables.

PiperOrigin-RevId: 729190812

Author: tensorflower-gardener

site/en/guide/versions.md

@@ -87,7 +87,7 @@ repository than TensorFlow.
 Because of this, we use a different version number for TensorFlow Lite
 (`TFLITE_VERSION_STRING` in `tensorflow/lite/version.h`, and `TfLiteVersion()`
 in `tensorflow/lite/c/c_api.h`) than for TensorFlow (`TF_VERSION_STRING` in
-`tensorflow/core/public/version.h`, and `TF_Version()` in
+`tensorflow/core/public/release_version.h`, and `TF_Version()` in
 `tensorflow/c/c_api.h`). Currently, these two version numbers happen to have the
 same value. But in future, they may diverge; for example, we may increment the
 major version number for TensorFlow Lite without incrementing the major version
@@ -369,7 +369,9 @@ guidelines for evolving `GraphDef` versions.
 There are different data versions for graphs and checkpoints. The two data
 formats evolve at different rates from each other and also at different rates
 from TensorFlow. Both versioning systems are defined in
-[`core/public/version.h`](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/public/version.h).
+[`core/public/version.h`](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/public/version.h)
+and
+[`core/public/release_version.h`](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/public/release_version.h).
 Whenever a new version is added, a note is added to the header detailing what
 changed and the date.