First: I checked my AD&D1 DMG, and indeed, a PC should stop gaining xp immediately at gaining the minimum for the next level. It is actually typed in all caps, but at the very end of the 2.5 pages of xp info (and you remember the density of the DMG’s text

I was remembering the rule from the BD&D rules, that xp stops just short of the second level.
Personally, I never knew anyone who used this rule -- I wouldn’t use it today, either. But then I also still wouldn’t use the training rules either. Such rules in AD&D1 really just made xp and leveling logistics too tedious and aggravating.
The thief says, “Hey, everyone, let’s take a break from adventuring for a few weeks so I can train up to my new level.”
The magic-user says, “No way, I’m not even half way to my next level yet.”
The fighter says, “If we take a break now, the BBEG will just pack up and leave his lair; we’ll never finish this adventure if we wait for you to train up.”
The cleric says, “Just suck it up. We’ll all take a break after finishing this, and we’ll all level up together.”
The paladin says, “Besides, you don’t have enough gold yet to pay for your training.”
The thief says, “Crap. A fat lot of good it does me to have low xp requirements for each level. I might as well be playing a barbarian.”
I wonder if you factored in the wasted XP from the moathouse and had the characters start the temple proper at level 2 (except for the thief at level 3) if that might not put the party right about where I would expect them to be (level 6-7).
No offense intended, but I’m not going to massage the data to get it to show what you expect it to show, or to reflect your personal experiences with the adventures. We have no evidence that your personal expectations or experience is the norm or intention for the adventures. You can consider the levels calculated with the data to be the maximum potential of these adventures.
On the topic of "efficiency" is there a general assumption in D&D3 that a party will kill every possible bad guy and collect every possible piece of treasure in an adventure? Because this certainly isn't the case in AD&D1 (or at least most AD&D1 modules) -- there are red herring encounters designed to waste resources that wise players will try to avoid*, there are 'treasures' that do the same (a party that loads themselves down to the 3" move class trying to glean every last copper and silver piece is setting themselves up to be waylaid by wandering monsters on their way back to the surface), plus 'easter egg' treasures that only the luckiest or most thorough (which, again, is a trade-off -- the longer you spend searching for hidden treasure the more wandering monsters you'll face) parties will ever find. Given all this, 75% efficiency of monsters slain and treasure recovered is probably about the best that can be expected, and the other 25% is just there to tempt the greedy or incautious into overreaching...
There is no “efficiency” assumption in any edition of this game (with regards to “clearing out” a dungeon).
There are many variables in all these adventures that can result in different end levels, treasure, and magic for the PCs. In the Moathouse, if the PCs fight their way past the six 1st-level fighters and one 2nd-level fighter, and then have a hard time with the 4th-level fighter, they may have to retreat out of the dungeon for a day or so of rest and recuperation. When they go back in, they may find that Lareth has taken all his treasure and left the lair. In such a case, the PCs loose out on a lot of treasure and xp. This could happen with any party in any edition. But this data collection (the adventure module itself) assumes the PCs will encounter Lareth and defeat him. (EGG even tells the DM what to do/what will happen after the PCs kill Lareth – an assassin will come after the PCs.)
Or, in the time the PCs are out, Lareth might bring in/hire more guards and humanoids. When the PCs come back and finish the adventure, they may actually gain more xp than was originally placed in the dungeon.
It's stated explicitly in the Moldvay-edit Basic Set (1981), and probably in the Mentzer-edit Basic Set (1983) as well that characters should receive approx. 80% of their XP from treasure and only 20% from monsters.
Where is this in the Moldvay set?
XP for Gold awards in AD&D were likewise reduced - using a different formula, IIRC - based on the difficulty of the fight.
What about treasure found without fighting enemies to get it?
So, I will continue listing the data as I have. Polls here on ENWorld show that most people never used the “measure of challenge” or the training rules in actual AD&D1 play. (I also have read that Gygax, himself, did not use those rules – MerricB, do you have a link to those statements?) Although polls on ENWorld shouldn’t be taken as universal fact, it does show me that most people here find the data in line with their real/actual play experiences.
Quasqueton