Use line buffering by awk in make import-sql

[zstadler]

When running quickstart.sh, the output is redirected to a pipe.
When the output is not a terminal, awk uses block buffering.
As a result, the output from make import-sql is block-buffered and the progress is not clear, especially when dealing with a large data set or a less powerful machine.

For example, here is a snapshot of the output:

CREATE FUNCTION
Time: 1.935 ms
CREATE TRIGGER
Time: 2.219 ms
psql:/sql/parallel/water__wate

The last line is completed only when the next buffer is flushed:

CREATE FUNCTION
Time: 1.935 ms
CREATE TRIGGER
Time: 2.219 ms
psql:/sql/parallel/water__waterway.sql:525: NOTICE:  trigger "trigger_store" for relation "osm_waterway_linestring" does not exist, skipping

The issue can be resolved by replacing

import-sql: all start-db-nowait
	$(DOCKER_COMPOSE) run $(DC_OPTS) openmaptiles-tools sh -c 'pgwait && import-sql' | \
	  awk -v s=": WARNING:" '$$0~s{print; print "\n*** WARNING detected, aborting"; exit(1)} 1'

with

import-sql: all start-db-nowait
	$(DOCKER_COMPOSE) run $(DC_OPTS) openmaptiles-tools sh -c 'pgwait && import-sql' | \
	  awk -v s=": WARNING:" '1{print; fflush()} $$0~s{print "\n*** WARNING detected, aborting"; exit(1)}'