Initiative has no real effect on resolving this.
Initiative only indicates who resolves effects on his objects first.

Lets check if I got that right.
Mage A has a creature (2HP, just for a change) under Mordok's Obelisk.
Mage B has a Death Link with a creature.

Mage B has an Initiative. He decides that he will resolve *himself* first, and heals 2HP, placing 2 damage on a Deathlinked creature.
Now Mage A resolves two effects on his creature: 2 direct damage and the loss of a creature (because he refuses to pay upkeep). What are his options? Common sense hints that he cannot destroy his creature first, somehow "cancelling" his opponent's healing. It it correct?

If Mage A has the Initiative, he starts with destroying his creature. Mage B cannot resolve Death Link... too bad for him.

Is it correct?

"You always choose the order in which events that affect your creatures and objects occur during this phase". So, always. And in Official Rules Supplement no one example which can break this rule.
But we have example for the second part: "In the rare case that a timing issue occurs, the player with the initiative decides the order."

Essentially that's the problem: these two phrases are somewhat contradictionary. Always except rare cases? What are these rare cases, how should one identify these? Why, for example, Death Link vs Regeneration is considered to be a timing issue (if I got it right), and Ghoul Rot vs Reneneration is not?

I'd say we have a good candidate for a new section in the next Official Rules&Codex Supplement.

no 2 time the same enchantment on the same object!

Ouch... Such a dumb and obvious mistake. Of course.

Problem is when its the other way around. THEN initiative matters.

If warlock has deathlink on a highland unicorn with regen that has 2 life left.
When the warlock has initative he can choose to resolve his deathlink before the horse gets to regen.

That makes perfect sense, thank you!

The key concept is that the Paladin has LoS to the zone that Steep Hill is attached to. Thus, he can build walls on any side of that zone.

Exactly what I said when discussing the problem with TS in a local chat. However it seems things have changed.
Rulebook (recent version available for download): "At  least one of the two zones adjacent to the target zone border must be in range, and the zone border in LoS when you cast the wall spell"
Old rulebook I've downloaded previously: "At  least one of the two zones adjacent to the wall must be in range and LoS when you cast the wall spell"

So strictly speaking, building a wall requires a LoS to the zone border, not the zone.

Bloodfire Helmet is listed with "Foo" subtype.