I'm asking about the practice of using the book as a measuring rod to decide if other official content in the system is balanced in comparison.
Core is a bare-minimum standard, but not a good standard
You absolutely do not want anything more powerful than the most-powerful things in core, or something less powerful than the least-powerful things in core.
However, the core is so wildly imbalanced that this is pretty difficult to do in the first place, making core a less-than-useful metric to standardize against. The Core Rulebook is nearly-identical to the D&D 3.5e core’s balance, and the 3.5e core was hideously imbalanced.
In particular, core has systemic problems between magical and non-magical classes. The more magical a class is, the more powerful it is. If you only compare things against comparable things (spells against spells, feats against feats, classes against similar classes), you will replicate that problem.
Supplements have somewhat improved the balance, and offer better guides
Consider the tiers of Pathfinder classes—you will notice three of the core classes (cleric, druid, wizard) are in the top tier, while two (monk and rogue) are in the bottom tier. Only one class (bard) actually lands in the middle tier, and even that is dubious (as nerfs to the bard relative to 3.5e make the bard stuggle to keep pace there).
Meanwhile, in Pathfinder as a whole, the middle tier is actually the most-heavily represented. Alchemist, hunter, inquisitor, investigator, magus, mesmerist, occultist, skald, spiritualist, and warpriest, along with the unchained versions of monk, rogue, and summoner, represent a huge chunk of the game, but none of them are in core.
Which basically is to say that the Core Rulebook is representative of the highs and lows of the entire Pathfinder system. Few classes more powerful than, say, wizard were ever published (only arcanist has much claim there), while few classes weaker than, say, monk were ever published (kineticist, possibly shifter).
So you definitely don’t want to fall outside that wide range, but you also would like for a tighter range than that.
Consider 6-level spellcasters
Ultimately, Pathfinder does best when it is dealing with 6-level spellcasters. This hits a nice sweet spot of classes that have plenty of options and ways of “cheating” (because that’s what magic is, and why magic is always best in this game—non-magical options aren’t allowed to cheat), but without being able to cheat too much (because the highest-level spells are patently absurd). But Core only had 1 of them, and one of the worst of them (unlike most, who also have another strong gimmick to pair with their spells, bardic performance is painfully limited).
These also serve as a useful metric for judging the balance of non-class options: if you can determine what level some option becomes available, you can determine whether or not it’s the kind of thing that would be level-appropriate for one of these better-balanced classes. If a feat replicates the effect of some spell they’ve been able to cast for 3 levels, it’s a pretty poor feat (unless it has some other major advantage). If a spell completely replaces a higher-level mundane trick, it’s probably a problematic spell. And so on.