Well, to use your terms:
(For the sake of brevity, I will be using acronyms; henceforth, AL is defined as "Axis of Legality", AM is "Axis of Morality", and AI is "Axis of Intent". Additionally, each category is to be understood as being from a player's perspective - they would, by necessity, be different from the perspective of the DM or the people who wrote the game.)
1. The game rules themselves act on the AL (as a "code of law").
2. The DM acts on both the AM (as upholder of the game's social construct) and on the AL (as arbiter of the rules).
3. The players act on both the AI (as arbiter of their character's actions) and the AM (as upholders of the game's social construct).
OOC problems arise when someone either fails to act on one of the axes OR acts on an axis they aren't supposed to act on. For example, a rules-lawyer is a player that acts on the AL, a DM is railroading when they act on the AI, a rule is bad if it fails to act on the AL in a way that makes sense, a player is an jerk if they fail to act on the AM, a DM is poor if they fail to act on the AL, and so on.
This is not to say that acting on a "wrong" axis is a route to guaranteed failure; any party that already acts on that axis can extend access to a different party (asking the DM or a fellow player to help you make a decision, the DM asking one of the players to be in charge of a subset of rules, and so on and so forth.)
An interesting observation is that a game doesn't actually need to have all three of the above roles; it just needs to have at least one party acting on each axis. So you could remove 1 (having a game without codified rules) or 2 (having a game without a DM); while this might require some rejiggering, it is certainly doable. However, removal of 3 causes the game (at least, as a group exercise) to fail miserably; playing a "solitaire" game simply requires the single player to act as both 2 and 3 (or requires using a pre-generated aid to "act as" 2).
Another observation is that a healthy game dynamic "filters" through the levels; if a player wants a rule to change, they approach the DM. Similarly, if the DM or a player has an issue with an in-game choice another player made, they approach them. It is vital to note that, while each party can only act on 1~2 axes, they can be acted ON through all of them.
So, to answer your "stuff to think about":
1. The first question is already covered by the role of the DM as an arbiter of unclear or undesired rules; they act on the proper axis. Now, if they revise a rule in such a way that it annoys or offends a player, that player can act on the AM to get them to reconsider. (So that's a "Yes, with the caveat that the 'local law' is subject to critique by the party.")
2. The second question is actually nonsensical; any ruling made by such a panel of experts would, logically, fall in category 1 above, and would therefore be superseded by the DM (unlike in an actual legal system, where the categories are divided very differently). It's equivalent to asking "Should the rules be well-written in the first place?" (Also, how would such a panel of experts be picked?)
3. The third question is, I think, based on an erroneous belief that "lawyer English" and "English" are conceptually distinct. You see, when a lawyer argues a technicality, what they are actually doing is arguing that the intent of the law does not cover that situation. This is why precedent is so important in legal systems; it helps the judge determine whether the intent of the law is applicable in this situation (which in turn tweaks the future applicability, and so on and so forth.)
In the microcosm of the game, the rules should be read in a way that gives you the most reasonable intent. If a rule is well-written, the intent will clearly flow from the text itself; if a rule is poorly written, and produces a nonsensical or ambiguous "RAW", interpretation is required. Now, at the table the DM works to pick up any slack left by "bad" rules; however, on forums such as these we have to invent a DM "system" that can pull off that arbitration for us.
Generally, we design one that reads the rules literally (with some small, agreed-on rule tweaks) - this causes the breakdown we refer to as TO, where rules that would normally be revised or disallowed are applied to devastating effect (You can actually do this with any sufficiently heavy system; D&D is just a game whose rules have been the most heavily dissected.)
Some other little tidbits that I couldn't fit in the body of the post:
Wow, RPGs are very much a theocracy; you've got one or two guys handling both the laws and arbitrating moral right and moral wrong. Huh.
Rule 0 is not technically a fallacy; it's only a bad idea because it overloads the DM, and because it reflects a failure to act on the AL on the part of the game rules. So bad idea, not fallacy.
"But wait, aren't players supposed to know the rules? Otherwise, the game gets so bogged down." and "What, players aren't allowed to know the rules? WTF!" are some responses I expect to get due to how I categorized things. I would just like to say right now that my categories only cover who has the ability to decide stuff on a given axis.
Ideally, players should know at least the rules necessary to play their character, just like how a private citizen should have an idea of what laws apply to them - however, they don't have the ability to decide whether something is rules-legal or not. Similarly, the DM should be aware of what any given player wants to do in-game, but can't decide what their character does without some permission.
Yes, this makes rules that allow the DM to claim a player's character by Imminent Domain (Imminent Actor?) bad unless they are ultimately an outcome of player intent. This applies to both removing a character from the game (killing someone's character) and abruptly converting a PC into an NPC.