Extend unit tests for portable executor runner input handling and add support for remaining input types

Some new features have been added the the executor runner, which is currently used by Arm backend so indirectly tested. However any specific executor runner unit tests have not been updated.

https://github.com/pytorch/executorch/blob/main/examples/portable/executor_runner/executor_runner.cpp#L53
So new features without a unit tests are following options,
DEFINE_string(inputs, "", "Comma-separated list of input files");
DEFINE_string(
output_file,
"",
"Base name of output file. If not empty output will be written to the file(s).");

DEFINE_bool(
print_all_output,
false,
"Prints all output. By default only first and last 100 elements are printed.");

Additionally in combination with above -input flag support for non tensor inputs have been added.
https://github.com/pytorch/executorch/blob/8da822cd154db854eade5562a28608b964558d17/extension/runner_util/inputs.cpp#L81

Definition of done: Update all possible non tensor inputs (tag values) in inputs.cpp - see EXECUTORCH_FORALL_TAGS. Add relevant unit tests in extension/runner_util/test/inputs_test.cpp.

For more context see https://github.com/pytorch/executorch/pull/14377

cc @mergennachin @iseeyuan @lucylq @helunwencser @tarun292 @kimishpatel @jackzhxng

Read examples/portable/executor_runner/executor_runner.cpp and extension/runner_util/inputs.cpp, then inspect extension/runner_util/test/inputs_test.cpp. Run the existing input tests first; done means inputs.cpp handles all tags listed by EXECUTORCH_FORALL_TAGS and relevant unit tests cover them, along with the runner input-related options.