The difficulty in Souls games is vastly overrated, but this is simply because the majority of the game's core audience (i.e. younger players) had only been exposed to extreme, degenerate levels of hand-holding in videogames, so much so that the defining feature of the medium - challenge, and the overcoming of challenge - was nearly completely forgotten by 'mainstream' games at the time.
I agree that dying in Souls only has short-term consequences and that it's possible to grind and eventually trivialize everything to a point, but you can still be killed by regular enemies in the level if you're not paying attention, no matter how much time you spend. It was the first time these newer players were exposed to a system that didn't simply take time as a substitute for skill - it might seem tautological to older players, but a game that requires you to actually learn through trial and error and overcome challenges is a rare thing these days.
So, from a utilitarian, Stuart Mill-esque perspective, the net result of the Souls series is good, because it revitalized the mainstream of the medium with a crucial aspect I'd even call it axiomatic - that had been forgotten: the necessity to overcome challenges through skill, coupled with a reasonable tradeoff between risk and reward (even if that risk is, on final analysis, almost nonexistent due to respawning enemies, checkpoints and potential endless repetition).
It's absurd to even think that this notion needed to be revitalized, but such are the times we live in. There are people everywhere who try to justify the existence of games as art, as narrative, as whatever else outside of the realm of gameplay. I define games simply by one aspect: challenge, and the careful balancing act between dedication and payoff.
The fact that people found again that you could feel exhilarated and ecstatic about beating a boss, or reaching the end of a tough level by the skin of your teeth, is great. It's a feeling that never gets old and that many games of my childhood gave me. It's something that 'modern' games have largely robbed their audience of - tilting exclusively on the side of reward, showering players with nothing but positive reinforcement for the most banal, menial tasks - therefore rendering their existence futile.
On the topic of saving/reloading: the issue of savescumming has existed for me for as long as games have allowed me to save whenever I want. The earliest notable example I can think of is Doom, where I realized that it was much more rewarding to go through the levels without dying and without previous equipment instead of saving/reloading every time you die.
However, Doom was a very well-designed game that supported that style of play explicitly - Romero has said all levels were explicitly designed to support a pistol start. Not all games can strike that careful balance, and most of them opt out of even trying.
I generally think choice is a good thing, but I'm not against limitations on saving. I usually prefer to play on the hardest difficulty level and only use one save slot with minimal backtracking (if a non-critical PC dies during a fight in a RPG I won't reload but will instead try to play the fight out, etc.)
I'm actually a fan of the original Wizardry style of saving, even if it was abusable (I didn't abuse it at the time). But that's just me. I'm more motivated to think scenarios through and optimize my style if there are real opportunity costs to failing.
Other people just want to relax, and that's fine too. That's why I think choices are good. But choices should never supplant the grueling work that goes into actually designing games to be challenging without being unfair. Unfortunately, most 'game designers' today are not engineers and do not think problems through - game design is a very top-down affair, just like many other fields of human creativity. The band-aid 'solution' to this problem is input from the players, which often makes matters worse.