The point you're trying to make is a valid one, but your expression of it isn't technically correct.  Miners can easily fake their 'support' for any flag, and they often have in the past-- e.g. signaling BTC1 when they weren't running it and had zero intention of ever running it.

So while it's true that miner signaling is sybil resistant,  the mining pools signaling flags don't necessarily have much skin in what they signal, so fairly little incentive to do so honestly.  This is particularly true because the parties with an investment in hardware are hashers (whom don't control the flags) and not pools (who do control the flags.  Many pools have significant altcoin investments and it's not too hard to imagine a kickback or altcoin play making fake signaling in a pool's interest.  Even hashers can move hashpower to altcoins that might gain value if Bitcoin had issues.

You could imagine a scheme where people pay _ethereum_ into a black hole to vote for bitcoin proposals. ... it would be an unfakable measurement, but if anything it would be anti-correlated with the wants and best interest of the Bitcoin community.

More generally,  we should always be careful when we're substituting what we can measure with what we need.  Measuring hashpower is probably only a reliable proxy when people haven't realized that they can game it, and we know for a fact now that they have.

The real unfakable metric is the eventual market price of the resulting asset -- but we can't know that in advance, only make educated guesses.