Problem with CSV export fom Google Earth Engine: final table is not what I want

I am trying to download a CSV file with a time series of Landsat 8 EVI2, and with the respective date and plot id from multiple points using GEE according to the following code:

var roi = ee.FeatureCollection([
 ee.Feature(ee.Geometry.Point([-43.8115,-16.6648]), {plot_id: 1}),
 ee.Feature(ee.Geometry.Point([-43.7923,-16.663]), {plot_id: 2}),
 ee.Feature(ee.Geometry.Point([-43.7794,-16.6596]), {plot_id: 3}),
 ee.Feature(ee.Geometry.Point([-43.8879,-16.7316]), {plot_id: 4}),
 ee.Feature(ee.Geometry.Point([-43.8487,-16.6756]), {plot_id: 5}),
 ee.Feature(ee.Geometry.Point([-43.8274,-16.6888]), {plot_id: 6}),
 ee.Feature(ee.Geometry.Point([-43.8946,-16.7579]), {plot_id: 7}),
 ee.Feature(ee.Geometry.Point([-48.3,-18.9833]), {plot_id: 8}),
 ee.Feature(ee.Geometry.Point([-44.1224,-16.3708]), {plot_id: 9}),
 ee.Feature(ee.Geometry.Point([-44.1134,-16.3687]), {plot_id: 10})
]);

// Intervalo de tempo da minha série temporal
var startdate = '2013-04-01'; 
var enddate = '2024-04-30'; 

var years = ee.List.sequence(ee.Date(startdate).get('year'), ee.Date(enddate).get('year'));


// This function masks clouds in Landsat 8 imagery.
 function maskL8(im) {
   var qa = im.select('QA_PIXEL');
   var mask = qa.eq(2720);
   return im.updateMask(mask).copyProperties(im);
 }

var ls8toa = ee.ImageCollection('LANDSAT/LC08/C02/T1_L2')
.filterBounds(roi)
  .filterDate(startdate, enddate)
  .filter(ee.Filter.lt('CLOUD_COVER',0.3))
    .map(function(im) {return maskL8(im)});

// Calcula EVI2 a partir do Landsat8

var ls8_evi2 = ls8toa.map(function(image) {
  var evi2 = image.expression(
    '2.5*(NIR-RED)/(NIR+2.4*RED+1)', {
      'NIR': image.select('SR_B5'),
      'RED': image.select('SR_B4')
}).rename('EVI2');
return image.addBands(evi2);
});

var landsat = ee.ImageCollection(ls8_evi2);
var EVI2 = ls8_evi2.select(['EVI2']);
print(EVI2, 'EVI2');
print('Quantas Imagens', EVI2.size());

//Função pra calcular valores médios para cada ponto
var pointsmean = function(image) {
  var means = image.reduceRegions({
    collection: roi.select(['plot_id']),
    reducer: ee.Reducer.mean(),
    scale: 30
  });
  
  // Atribuindo datas
  means = means.map(function(f) { return f.set({date: image.date().format("YYYY-MM-dd")}) });
  
  return means.copyProperties(image);
};

var finalEVI2 = EVI2.map(pointsmean).flatten()
.sort('date', false);

print(finalEVI2, 'final EVI2');

// Exportar tabela em csv direto pro Drive
Export.table.toDrive({
collection: finalEVI2,
  description: 'EVI2_walter_'+startdate+'_TO_'+enddate,
  folder: 'LS8_walter',
fileFormat: 'CSV'
});

Actually, I used 34 pairs of coordinates instead of 10 as shown in the code above. I simplified the sample in this Question.

The output shown in the console is:

The CSV file is not what I expected (see figure below). I need columns named ‘date’, ‘plot_id’, and ‘EVI2’ with the values.

What is wrong and how to fix it?