Print query detail when client uses -f or --execute with --debug option

[ebyhr]

Previously, presto cli with -f or --exceucte only shows result set even if --debug option is used.
By this commit, presto cli prints the details the same as interactive mode.

-f and --debug

$ cat test.sql
select * from tpch.sf10000.customer limit 1;

$ presto --debug -f test.sql
"750000001","Customer#750000001","LC4O0BHw2 wNroftaCEp0SDGZ0r1SQV4nm4Gt","11","21-456-231-7942","-765.35","HOUSEHOLD","s wake carefully outside the regular instructions. ironicall"

Query 20190112_063454_00067_x3xmi, FINISHED, 1 node
http://localhost:8080/ui/query.html?20190112_063454_00067_x3xmi
Splits: 21 total, 20 done (95.24%)
CPU Time: 0.1s total,  125K rows/s,     0B/s, 70% active
Per Node: 0.9 parallelism,  107K rows/s,     0B/s
Parallelism: 0.9
Peak Memory: 0B
0:00 [16.4K rows, 0B] [107K rows/s, 0B/s]

--execute and --debug

$ presto --debug --execute "select * from tpch.sf10000.customer limit 1;"
"1125000001","Customer#1125000001","Uj6zjz2geB9Y9mDZlmgiEGvr3","23","33-135-417-9700","-180.05","HOUSEHOLD","l instructions. carefully special"

Query 20190112_063519_00068_x3xmi, FINISHED, 1 node
http://localhost:8080/ui/query.html?20190112_063519_00068_x3xmi
Splits: 21 total, 21 done (100.00%)
CPU Time: 0.1s total,  156K rows/s,     0B/s, 88% active
Per Node: 1.0 parallelism,  162K rows/s,     0B/s
Parallelism: 1.0
Peak Memory: 0B
0:00 [21.9K rows, 0B] [162K rows/s, 0B/s]

No changes if --debug isn't used.
-f

$ presto -f test.sql
"1125000001","Customer#1125000001","Uj6zjz2geB9Y9mDZlmgiEGvr3","23","33-135-417-9700","-180.05","HOUSEHOLD","l instructions. carefully special"

--execute

$ presto --execute "select * from tpch.sf10000.customer limit 1;"
"375000001","Customer#375000001","A2qPnPHny3EfSYRt5HRK xbO25","14","24-835-946-7501","873.28","HOUSEHOLD"," dogged requests. packages are bold"

[nezihyigitbasi]

With this change the query details will be printed to stdout, which is usually not a good idea. -f is usually used for batch runs/scripting and the output (which was previously only data) is usually redirected to a file. Now the output file will contain both data + stats, etc., which will break existing scripts and also will make difficult to pipe stuff together with this option (as the output will be csv/tsv + some other unstructured text, the query details).

Posix defines these streams as "standard output (for writing conventional output), and standard error (for writing diagnostic output)", so I think in -f mode we should write these details to stderr.

[ebyhr]

@nezihyigitbasi Thank you for your comment. Let me fix the output direction to stderr.

[ebyhr]

Changed output stream to stderr.

$ presto --debug -f test.sql 2>/dev/null
"7500001","Customer#007500001","o NIcrMHVcnYy3b6Or","24","34-544-477-2141","4996.81","BUILDING","ully alongside of the deposits. even, even f"
$
$ presto --debug -f test.sql
"11250001","Customer#011250001","RNFodMnae9vkA5Bswzb42rnp","14","24-397-921-1461","4930.66","HOUSEHOLD"," along the even packages nag according to the "

Query 20190115_132958_00012_vncji, FINISHED, 1 node
http://localhost:8080/ui/query.html?20190115_132958_00012_vncji
Splits: 21 total, 21 done (100.00%)
CPU Time: 0.1s total,  159K rows/s,     0B/s, 76% active
Per Node: 0.9 parallelism,  141K rows/s,     0B/s
Parallelism: 0.9
Peak Memory: 0B
0:00 [21.9K rows, 0B] [141K rows/s, 0B/s]

$ presto --debug --execute "select * from tpch.sf10000.customer limit 1;" 2>/dev/null
"1","Customer#000000001","IVhzIApeRb ot,c,E","15","25-989-741-2988","711.56","BUILDING","to the even, regular platelets. regular, ironic epitaphs nag e"```
$
$ presto --debug --execute "select * from tpch.sf10000.customer limit 1;"
"750000001","Customer#750000001","LC4O0BHw2 wNroftaCEp0SDGZ0r1SQV4nm4Gt","11","21-456-231-7942","-765.35","HOUSEHOLD","s wake carefully outside the regular instructions. ironicall"

Query 20190115_133029_00013_vncji, FINISHED, 1 node
http://localhost:8080/ui/query.html?20190115_133029_00013_vncji
Splits: 21 total, 21 done (100.00%)
CPU Time: 0.1s total,  153K rows/s,     0B/s, 67% active
Per Node: 0.9 parallelism,  145K rows/s,     0B/s
Parallelism: 0.9
Peak Memory: 0B
0:00 [21.9K rows, 0B] [145K rows/s, 0B/s]