`env` attribute of `*_binary` rules not applied when used via `--run_under`

### Description of the bug:

Example:

```
sh_binary(
name = "my_script",
srcs = ["my_script.sh"],
env = {"FOO": "BAR"},
)
```

If I do `bazel test --run_under=//:my_script //:my_test`, then the `env` attribute of `my_script` is not used.

I would somehow expect that it does, because the script is being `run`.

This is particularly needed when the script needs to access the location of some `data` to function correctly:

```
sh_binary(
name = "my_script",
srcs = ["my_script.sh"],
data = ["@some//:data"],
env = {"DATA_LOCATION": "$(rlocationpath @some//:data)"},
)
```

### Which category does this issue belong to?

Core

### What's the simplest, easiest way to reproduce this bug? Please provide a minimal example if possible.

_No response_

### Which operating system are you running Bazel on?

Ubuntu 24.04

### What is the output of `bazel info release`?

8.6.0

### If `bazel info release` returns `development version` or `(@non-git)`, tell us how you built Bazel.

_No response_

### What's the output of `git remote get-url origin; git rev-parse HEAD` ?

```text

```

### If this is a regression, please try to identify the Bazel commit where the bug was introduced with bazelisk --bisect.

_No response_

### Have you found anything relevant by searching the web?

_No response_

### Any other information, logs, or outputs that you want to share?

_No response_

Reproduce the issue with the shown sh_binary and bazel test --run_under command, then compare the environment seen by the wrapper with the environment from a direct run. Trace how --run_under launches the binary and how its env and data attributes are propagated; done means FOO and DATA_LOCATION are available when the wrapper runs.

Written by the indexing model from the issue text.