specifies what part is covered by the tensor to save.

`slice-spec` itself is a `:`-separated list: `slice0:slice1:...:sliceN-1`
where each `sliceI` is either:

*  The string `-` meaning that the slice covers all indices of this dimension
*  `start,length` where `start` and `length` are integers.  In that
   case the slice covers `length` indices starting at `start`.

See also `Save`.

                  `"*.example.com"`, and `"*.example.com"` takes precedence over `""`.


                  Additionally, if there are multiple wildcard entries, more specific
                  wildcard entries must be processed before less specific wildcard entries.
                  For example, `"*.foo.example.com"` takes precedence over `"*.example.com"`.

		{"BenchmarkResult.MemBytes", Field, 1},
		{"BenchmarkResult.N", Field, 0},
		{"BenchmarkResult.T", Field, 0},
		{"Cover", Type, 2},
		{"Cover.Blocks", Field, 2},
		{"Cover.Counters", Field, 2},
		{"Cover.CoveredPackages", Field, 2},
		{"Cover.Mode", Field, 2},
		{"CoverBlock", Type, 2},
		{"CoverBlock.Col0", Field, 2},
		{"CoverBlock.Col1", Field, 2},

33A6          ; mapped                 ; 006B 006D 0033 #1.1  SQUARE KM CUBED
33A7          ; mapped                 ; 006D 2215 0073 #1.1  SQUARE M OVER S
33A8          ; mapped                 ; 006D 2215 0073 0032 #1.1 SQUARE M OVER S SQUARED
33A9          ; mapped                 ; 0070 0061     # 1.1  SQUARE PA
33AA          ; mapped                 ; 006B 0070 0061 #1.1  SQUARE KPA

        `is_dynamic_op=False` would be set.
    *   `converter.convert()` no longer returns a `tf.function`. Now the
        function must be accessed from the saved model.
    *   The `converter.calibrate()` method has been removed. To trigger
        calibration, a `calibration_input_fn` should be provided to
        `converter.convert()`.

*   Other: