The paper Extreme events reveal an alimentary limit on sustained maximal human energy expenditure, Thurber et al (2019) answers a related but slightly different question, namely what ‘total energy expenditure’ (TEE) can humans sustain, as both a number of kcal, and as a multiple of Basal Metabolic Rate (BMR). That's the energy expended, as opposed to the calorific value of the food consumed.

There's plenty of detail there, but the paper concludes, inter alia,

• As the endurance event gets longer, the TEE per day decreases – ie, you can only keep this up for so long – and seems to converge towards a low-ish upper limit on the body's ability to convert food to activity, that can be kept up more or less indefinitely.
• And it indicates, more or less in passing, that the most hardcore, ultra-endurance human activity is to be pregnant for nine whole months.

From the discussion:

The relationship between SusMS and duration flattens out at ~2.5× BMR (Fig. 1B), suggesting a metabolic ceiling for habitual metabolic scope (often termed “physical activity level”) in humans. Prolonged expenditure above this metabolic ceiling (~2.5× BMR) requires consuming energy reserves and is not sustainable indefinitely.

That is, though humans can sustain much larger figures (~10 x BMR) for short periods, there seems to be a robust limit on how much they can sustain indefinitely, without losing weight. That appears to mean that, even if workers consume many calories, there is an alimentary system limit on how much energy the body can extract from this long-term. Figures 1b and 3b seem illuminating, here.

Or, focusing on the original question, it appears that, even if these frontiersmen managed to get 20 000 kcal/day down their throats, it doesn't follow that their guts managed to turn all of that into physical work, beyond an initial period (that is, and heading towards the scatological, they might have made a lot of dung beetles very happy).

Terms:

• Basal Metabolic Rate.
• ‘Total energy expenditure (TEE) is composed of the energy costs of the processes essential for life (basal metabolic rate (BMR), 60–80% of TEE), of the energy expended in order to digest, absorb, and convert food (diet-induced thermogenesis, ~10%), and the energy expended during physical activities (activity energy expenditure, ~15–30%)’ (from 10.1186/s40798-017-0076-1, which has further information about this sort of question).

Just one example I remembered (as the question-asker):

According to this documentary, Victorian industrial workers burned roughly 5,000 calories a day. This helped fuel a bread and bakery industry.

Certainly not for long

Take the Race Across AMerica for example. One of the most grueling endurance competitions in the world. This Case Report says the 4th place finisher in 2003 burned 15 100 - 23 280kcal per day (17 965kcal average). This would put him close to your 20Mcal estimate. However, that’s a modern world class athlete, doing it for 9 days with barely any sleep and certainly not something they or anyone else could keep up for long. Other fun facts: The athlete from the report “only” ate ~10Mcal per day and therefore lost 5kg of body weight in those 9 days. He also drank between 10 and 19l of water per day.

An “Iron Man” type triathlon competition burns 11 009kcal in 12.5 hours.

As for a sustainable energy consumption: World class long-distance cyclists ride up to ~700km/week during a year. Roughly estimated that’s 2000–4000kcal/day. If you add basal metabolic rate that’s less than 6000kcal/day.

I do think that cycling is a good exercise to pick for an upper boundary. I don’t think there is any other exercise humans can keep up for so long at such a high intensity. Running, rowing, hiking/mountaineering, cross country skiing or just general hard physical work (e.g. moving heavy things) might come close in perfect conditions.

Another limitation is the human digestive system. Eating 20Mcal in 16 waking hours would mean you have to eat 1250kcal every single hour. Even half of that is already a big challenge, especially when it’s not just simple carbs and fats.

A lot of information concerning the caloric intake of 18th century travelers in the Canadian wilds can be gleaned from a journal article published in Arctic Anthropology in 1993, "Always with Them Either a Feast or Famine": Living off the land with the Chipewyan Indians, 1791-1792, by June Helm. This can be read at JSTOR within the free monthly articles alloted. The article examines in detail the foodstuffs mentioned in the journals of Peter Fidler, a Hudson Bay Company employee who was 'embedded' with a group of Native Americans over the winter of 1791-92.

The abstract sums up what was the average consumption of the group by both weight and calories (the article goes into full details on what was consumed and how the calorie figures were arrived at)(emphasis mine):

ABSTRACT
Peter Fidler's record of food kills by his Chipewyan companions in the boreal forest south of Great Slave Lake through the winter of 1791/92 is combined with calculations of percentages of consumable tissue and kilocalories from the prey animals— bison, moose, and beaver— to yield a high-low range of edible tissues of 6.89-6.15 lbs per person-day and 5780-5140 kcal per person-day.

So this article gives some related information and shows that over a winter period the food intake for the party members was roughly 6.15-6.89 lbs. per day, with the calories come in at 5140-5780 kcal per person per day.

If we divide that out we get about 836 calories per pound of wild meat. At the "9 pounds of meat" per day ration allowance you mention from the original source, then we should expect your rowers were consuming roughly 7524 calories per day (though some variation depending on what game they consumed would of course be expected), not nearly the 20,000 from the title question.