Generate mosaic with the best NDVI of a month over a period of time

BestNDVIMosaic.js

/* 
 *  @notice GEE JavaScript for Code Editor
 *  @notice Generate a mosaic using pixels of the best NDVI of all images of a month
 *  @notice Do it over all same months of a period of time, for example March (or August, or December)
 *  @notice ... useful to remove clouds and calculate monthly NDVI stats over several years
 *  @author Vu Tien Khang, licence MIT, Jan 2021
 */
// make a list of years composing the history
var history = ee.List.sequence(2017, 2020);
// per year of history, calculate bestNDVI and add to bestNDVICollection
bestNDVICollection = history.map(makeBestCollection(studyMonth, refSetS2));
// do something with this collection...
//-------------------------
function makeBestCollection(studyMonth, refSetS2){
/*
 *  @notice Generate a mosaic using pixels of the best NDVI of all images of a month
 *  @param {integer in 1-12} studyMonth: month to generate ee.DateRange
 *  @param {imageCollection} refsetS2: full set of images, several years
 *  @dev {function} addNDVI: function to add NVI channel named 'NDVI' to an image
 */
return function(year){
    // ask GEE to set start of range of dates
    var mosaicStartDate = ee.Date.fromYMD(year, studyMonth, 1); 
    // ask GEE to set range of images to mosaic 
    var range2Mosaic = ee.DateRange(mosaicStartDate, mosaicStartDate.advance(1, 'month'));
    // ask GEE to keep only images in this range to mosaic 
    var filtered = refSetS2.filterDate (range2Mosaic);
    // ask GEE to add NDVI channel to all images of this filtered collection 
    var with_NDVI = filtered.map (addNDVI);
    // ask GEE to make a mosaic from all the set with the best NDVI and return this image
    return with_NDVI.qualityMosaic ('NDVI')
        .set('system:time_start', with_NDVI.first().get('system:time_start'));
  }
}
//---------------------------
function addNDVI(NIR, RED){ 
  /* 
  *   @notice Function used by map() to add to "image" the band NDVI (normalized difference NIR and Red)
  *   @param {ee.String} NIR: name of band NIR in image, depends on the satellite
  *   @param {ee.String} RED: name of band RED in image, depends on the satellite
  *   @author Vu Tien Khang, licence MIT, Jan 2021
   */
    return function (image) {
      var ndvi = image.normalizedDifference ([NIR, RED]).rename ('NDVI');
      return image.addBands (ndvi);
      }};

This snippet is a bit too "ninja" :-)
For beginners or students, I made also a complete ready-to-use GEE code: https://gist.github.com/kvutien/97be53cd7694d30ff06ddfaa742a399e