mdp chap clean up

[shensquared]

Notational/formatting stuff

• all the V and Qs wrapped inside \mathrm{}
• all Rewards R and transition T wrapped inside \mathrm{}
• horizon _h appears on the subscript, policy \pi or star on superscript
• horizon $h$ is lowercase
• abbreviation "MDP" should always be uppercase

Content stuff

(thanks Mardavij for comments too)

• openning paragraph
• 10.1.2 infinite-horizon
• 10.8 side note
• demote the DP notebox onto side note
• differentiate between fix-policy Q_{\pi} and optimal Q^*
• (in fact, typically small) remove
• decide if to introduce sink/terminal state