fix: pending x:expect should not trigger running SUT

[galtm]

I was doing some experiments with @catch to pinpoint the latest spot where catch="yes" needs to go if you want to avoid a hard error when running a test whose SUT produces an error. It seemed like a bug that a pending x:expect element caused the SUT to run. I expected a pending x:expect element not to influence test behavior at all.

This PR changes the logic for what causes the SUT to run from "this scenario isn't pending and has at least one x:expect child" to "this scenario isn't pending and has at least one x:expect child that also isn't pending."

Example of effect on test-driven development

On the master branch, the first scenario has a hard error, while the second scenario runs to completion.

  <x:scenario label="Function not written yet">
    <x:call function="function-not-written-yet"/>  
    <x:expect label="will someday return 3" select="3"
      pending="not written yet"/>
  </x:scenario>

  <x:scenario label="Function not written yet">
    <x:call function="function-not-written-yet"/>
    <x:pending label="not written yet">
      <x:expect label="will someday return 3" select="3"/>
    </x:pending>
  </x:scenario>

On the branch of this PR, both scenarios run to completion. The first scenario is similar to the catch.xspec addition in this PR.