Join ON problem + tolerance

**[Sorry for bad english]**
Here is my tickscript and I am trying to query from two differente measurements.

1. By **first query** getting the historical stats and then take the next monitor label with i would like to use as filter in the last join line.
2. By **second getting** the daily activation data for whole year period and then computing the statistics.

The first Join work fine but the second one based on the first don't.

I can't join the new stream with past stream or it is another problem for example. time tolerance or something else.

JSON file:
```
{
"template-id": "daily_monitor_template",
"dbrps": [{"db": "atsmi", "rp": "autogen"}],
"vars": {
"monitor": {"type": "int", "value": 1},

"statsDB": {"type": "string", "value": "atsmi"},
"statsRP": {"type": "string", "value": "autogen"},
"statsMeasurement": {"type" : "string", "value" : "devices_stats"},
"offsetStats": {"type" : "duration", "value" : "1h" },
"statsPeriod": {"type" : "duration", "value" : "5w"},

"cronStats": {"type" : "string", "value" : "0 3 1 * * *"},
"nextMonitor": {"type" : "string", "value" : "next_monitor"},

"inputDevicesDB": {"type": "string", "value": "atsmi"},
"inputDevicesRP": {"type": "string", "value": "autogen"},
"inputDevicesMeasurement": {"type" : "string", "value" : "devices_daily_activation"},
"dataPeriod": {"type" : "duration", "value" : "50w"},
"groupTime": {"type" : "duration", "value" : "1d"},
"cronData": {"type" : "string", "value" : "0 3 1 * * *" },

"deviceField": {"type" : "string", "value" : "event_counter" },
"offsetData": {"type" : "duration", "value" : "1h" },

"filterWhere": {"type": "lambda", "value": " \"filter_monitor.next_monitor\" == monitor"},

"every": {"type" : "duration", "value" : "15s"},
"deviceFilter": {"type": "lambda", "value": "\"id_device\" == '15576'"},
"statsFilter": {"type": "lambda", "value": "\"next_monitor\" == monitor"}

}
}

```
TICKscipt
```
// Testing variables
var every duration

var deviceFilter = lambda: TRUE
var statsFilter = lambda: TRUE
// COMMON VARIABLES
var monitor int

// decide which kind of data to filter -> depend on task file

// input stats vars
var statsDB string
var statsRP string
var statsMeasurement string

var cronStats string
var statsPeriod duration
var offsetStats duration

var groupStatsBy = [*]
var nextMonitor string

// input device daily data
var inputDevicesDB string
var inputDevicesRP string
var inputDevicesMeasurement string

var cronData string
var dataPeriod duration
var offsetData duration

var groupTime duration
var groupDataBy = *

var deviceField string
var filterWhere lambda

// query the next label from histrical stats data: take the last updated history for device and then filter for label
var filter_monitor = batch
|query('SELECT '+nextMonitor+' FROM "' + statsDB + '"."' + statsRP + '"."' + statsMeasurement + '"')
//.cron(cronStats)
.period(statsPeriod)
.groupBy(groupStatsBy)
// .offset(offsetStats)
.every(every)
.align()
|where(lambda: "next_monitor" == monitor)
|where(lambda: "id_device" == '15576')
|last(nextMonitor)
.as(nextMonitor)

// filter_monitor
// |log()

// // query the daily activation data from database
var device_data = batch
|query('SELECT SUM('+deviceField+') as '+deviceField+' FROM "'+inputDevicesDB+'"."'+inputDevicesRP+'"."'+inputDevicesMeasurement+'"')
//.cron(cronData)
.period(dataPeriod)
.every(every)
.groupBy(time(groupTime), groupDataBy)
.align()
// .offset(offsetData)
|where(deviceFilter)

// Calculation devices stats
var tot_activation = device_data
|sum(deviceField)
.as('sum')

var avg= device_data
|mean(deviceField)
.as('avg')

var std = device_data
|stddev(deviceField)
.as('std')

var min = device_data
|min(deviceField)
.as('min')

var max = device_data
|max(deviceField)
.as('max')

var spread = device_data
|spread(deviceField)
.as('spread')

var median = device_data
|median(deviceField)
.as('median')

var decimo_percentile = device_data
|percentile(deviceField, 10.0)
.as('p10')

var quinto_percentile = device_data
|percentile(deviceField, 5.0)
.as('p5')

var primo_percentile = device_data
|percentile(deviceField, 1.0)
.as('p1')

var device_stats = tot_activation
|join(avg, std, min, max, spread, median, decimo_percentile, quinto_percentile, primo_percentile)
.as('sum','avg','std','min','max','spread','median','p10','p5','p1')

// Does't work with ON, why ?
// .on('id_device')

.tolerance(5m)
.streamName('device_stats')
|eval(lambda: "avg.avg" - "std.std", lambda: "avg.avg" + "std.std")
.as('stdN', 'stdP')
.keep()
|eval(lambda: "sum.sum", lambda: "avg.avg", lambda: "std.std", lambda: "min.min", lambda: "max.max", lambda: "spread.spread", lambda: "median.median", lambda: "p10.p10", lambda: "p5.p5", lambda: "p1.p1")
.as('sum','avg', 'std', 'min', 'max', 'spread', 'median', 'p10', 'p5', 'p1')

var filter_data = device_stats
|join(filter_monitor)
.as('device_stats', 'filter_monitor')
.tolerance(5m)
//Don't work either with ON or tolerance, ??????????????????
// .on('id_device')
|log()
// |log()
// |default()
// .field('filter_monitor.'+nextMonitor, monitor)
// |delete()
// .field('filter_monitor.event_counter')
// |where(lambda: "filter_monitor.next_monitor" == monitor)

```

No source files or tests are named. Reproduce the supplied JSON variables and TICKscript, then compare the working first join with the later joins using ON and tolerance; done means identifying whether the join failure is caused by timestamps, grouping, or unsupported join behavior.

Written by the indexing model from the issue text.