Causal Effect Estimation with TMLE: Handling Missing Data and Near-Violations of Positivity