Does Binohash grinding behave like a PoW, and is it really ASIC-resistant?

I have been studying Robin Linus’s "Binohash" paper
(Binohash:transaction introspection with out softforks,https://delvingbitcoin.org/t/binohash-transaction-introspection-without-softforks/2288) and
making an attempt to grasp the grinding course of.

My tough understanding:

• The spender chooses subsets of signatures
  from a hard and fast pool embedded within the locking script
• Every subset modifications the scriptCode by way of
  FindAndDelete earlier than hashing
• The purpose is to discover a subset the place the ensuing
  sighash satisfies a leading-zero situation

So in follow this seems to be like a search drawback
over combinatorial subsets moderately than a easy
nonce area.

In comparison with Bitcoin mining:

• Mining hashes a fixed-size header with a
  altering nonce
• Binohash requires modifying KB-sized script
  knowledge (by way of FindAndDelete) earlier than every hash
• Every transaction defines its personal impartial
  search area

My questions:

1. Is it affordable to consider this as a sort
   of PoW, or is {that a} deceptive analogy?

2. What really limits ASIC acceleration right here —
   the FindAndDelete step, the bigger enter measurement,
   or one thing else?

3. Would a specialised ASIC nonetheless present
   vital speedup, or is that this inherently
   nearer to a GPU-style workload?

4. Does the truth that every transaction defines
   its personal search area make this basically
   non-competitive, not like mining the place everybody
   races on the identical puzzle?