Electrical – Why this high-pass filter behaves like a low-pass filter for high frequencies, on this ideal LTSpice simulation

The "Dual Decompensated" in the IC's title is biting you.

Two things come to mind. The description in the data sheet mentions that that IC is stable in gains of 10 and more, and you are low gain.

Also

From the data sheet: High Speed Operation When the feedback around the op amp is resistive (RF), a pole will be created with RF, the source resistance and capacitance (RS, CS), and the amplifier input capacitance (CIN ≈ 2pF). In low closed loop gain configurations and with RS and RF in the kilohm range, this pole can create excess phase shift and even oscillation. A small capacitor (CF) in parallel with RF eliminates this problem. With RS (CS + CIN) = RFCF, the effect of the feedback pole is completely removed.

Even though you have no resistor in your feedback path, maybe R2 is working its way in somehow. You might try a small cap in parallel with R2, but be careful of your frequency response. You could also try a different configuration filter.

Please remember that when using sim.okawa-denshi.jp it is up to you to know how to properly create the topology in question. It is ill advised to create a Sallen-Key topology with a gain greater than 2 because it tends to oscillate. In fact, the page itself even states (after typing in your numbers) that it will oscillate at a frequency of 401.949Hz. If you require more gain then please create another stage that focuses on that.

Edit: Even better, I noticed that you Q-factor and your Damping factor are both negative. That is a sure indication that you will have oscillatory behavior.

As a suggestion, please use the calculation that asks for fc, gain, and provide either the Q-factor or damping ratio. This will provide you with a more stable set-up (real op amps often require additional components for stability).