Component-separated, CIB-cleaned thermal Sunyaev--Zel'dovich maps from $\textit{Planck}$ PR4 data with a flexible public needlet ILC pipeline

We use the full-mission $\textit{Planck}$ PR4 data to construct maps of the thermal Sunyaev$--$Zel'dovich effect (Compton-$y$ parameter) in our Universe. To do so, we implement a custom needlet internal linear combination (NILC) pipeline in a Python package, $\texttt{pyilc}$, which we make publicly available. We publicly release our Compton-$y$ maps, which we construct using various constrained ILC ("deprojection") options in order to minimize contamination from the cosmic infrared background (CIB) in the reconstructed signal. In particular, we use a moment-based deprojection which minimizes sensitivity to the assumed frequency dependence of the CIB. Our code $\texttt{pyilc}$ performs needlet or harmonic ILC on mm-wave sky maps in a flexible manner, with options to deproject various components on all or some scales. We validate our maps and compare them to the official $\textit{Planck}$ 2015 $y$-map, finding that we obtain consistent results on large scales and 10-20$\%$ lower noise on small scales. We expect that these maps will be useful for many auto- and cross-correlation analyses; in a companion paper, we use them to measure the tSZ -- CMB lensing cross-correlation. We anticipate that $\texttt{pyilc}$ will be useful both for data analysis and for pipeline validation on simulations to understand the propagation of foreground components through a full NILC pipeline.