expect `Enable()` inTerm:'off' does not trigger failpoint code, but it's not as expect

Bug Report

• CODE(Transfromed)

func GetValue() int {
	if _, _err_ := failpoint.Eval(_curpkg_("failpoint-name")); _err_ == nil {
		fmt.Println("enable failpoint")
	}
	return 1
}

• TEST CODE

func TestGetValue(t *testing.T) {
	var err error
	fmt.Println("############case01###############")
	err = failpoint.Enable("test/failpoint-name", "return(true)->off")
	if err != nil {
		t.Fatal("enable failpoint failed:", err)
	}
	GetValue()

	fmt.Println("############case02###############")
	err = failpoint.Enable("test/failpoint-name", "off")
	if err != nil {
		t.Fatal("enable failpoint failed:", err)
	}
	GetValue()

	fmt.Println("############case03###############")
	err = failpoint.Enable("test/failpoint-name", "return(2)")
	if err != nil {
		t.Fatal("enable failpoint failed:", err)
	}
	GetValue()
}

• EXPECT:
  According to README, "off: Take no action (does not trigger failpoint code)", used to stop triggering of the failpoint. So my expectation is that case02 and case03 should not print "enable failpoint".
• ACTUAL:
  It seems no working. I'm not sure if I'm misunderstanding this argument.

go test -v -run TestGetValue
=== RUN   TestGetValue
############case01###############
enable failpoint
############case02###############
enable failpoint
############case03###############
enable failpoint
--- PASS: TestGetValue (0.00s)

Start with the README's definition of the off action and trace how failpoint.Enable handles return(true)->off, off, and return(2) in the transformed Go code. Run go test -v -run TestGetValue using the reported cases and compare whether the observed output matches the documented semantics. Done means the three cases produce the behavior described by the README.